Access to this full-text is provided by MDPI.
Content available from Journal of Clinical Medicine (JCM)
This content is subject to copyright.
Academic Editor: Lindsay A. Farrer
Received: 18 December 2024
Revised: 10 January 2025
Accepted: 14 January 2025
Published: 16 January 2025
Citation: Onciul, R.; Tataru,C.-I.;
Dumitru, A.V.; Crivoi, C.; Serban, M.;
Covache-Busuioc, R.-A.; Radoi, M.P.;
Toader, C. Artificial Intelligence and
Neuroscience: Transformative
Synergies in Brain Research and
Clinical Applications. J. Clin. Med.
2025,14, 550. https://doi.org/
10.3390/jcm14020550
Copyright: © 2025 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license
(https://creativecommons.org/
licenses/by/4.0/).
Review
Artificial Intelligence and Neuroscience: Transformative
Synergies in Brain Research and Clinical Applications
Razvan Onciul
1, 2, †
, Catalina-Ioana Tataru
3, 4,
*, Adrian Vasile Dumitru
1,5,6,
* , Carla Crivoi
7, †
, Matei Serban
1,8,9
,
Razvan-Adrian Covache-Busuioc 1,8,9, Mugurel Petrinel Radoi 1,8 and Corneliu Toader 1,8
1Department of Neurosurgery, “Carol Davila” University of Medicine and Pharmacy,
020021 Bucharest, Romania; razvan.onciul@drd.umfcd.ro (R.O.); matei.serban2021@stud.umfcd.ro (M.S.);
razvan-adrian.covache-busuioc0720@stud.umfcd.ro (R.-A.C.-B.); petrinel.radoi@umfcd.ro (M.P.R.);
corneliu.toader@umfcd.ro (C.T.)
2Neurosurgery Department, Emergency University Hospital, 050098 Bucharest, Romania
3Clinical Department of Ophthalmology, “Carol Davila” University of Medicine and Pharmacy,
020021 Bucharest, Romania
4Department of Ophthalmology, Clinical Hospital for Ophthalmological Emergencies,
010464 Bucharest, Romania
5Department of Morphopathology, “Carol Davila” University of Medicine and Pharmacy,
020021 Bucharest, Romania
6Emergency University Hospital, 050098 Bucharest, Romania
7Department of Computer Science, Faculty of Mathematics and Computer Science, University of Bucharest,
010014 Bucharest, Romania; crivoicarla02@gmail.com
8Department of Vascular Neurosurgery, National Institute of Neurovascular Disease,
077160 Bucharest, Romania
9Puls Med Association, 051885 Bucharest, Romania
*Correspondence: catalina-ioana.tataru@umfcd.ro (C.-I.T.); vasile.dumitru@umfcd.ro (A.V.D.)
†These authors contributed equally to this work.
Abstract: The convergence of Artificial Intelligence (AI) and neuroscience is redefining our
understanding of the brain, unlocking new possibilities in research, diagnosis, and therapy.
This review explores how AI’s cutting-edge algorithms—ranging from deep learning to
neuromorphic computing—are revolutionizing neuroscience by enabling the analysis of
complex neural datasets, from neuroimaging and electrophysiology to genomic profiling.
These advancements are transforming the early detection of neurological disorders, en-
hancing brain–computer interfaces, and driving personalized medicine, paving the way
for more precise and adaptive treatments. Beyond applications, neuroscience itself has in-
spired AI innovations, with neural architectures and brain-like processes shaping advances
in learning algorithms and explainable models. This bidirectional exchange has fueled
breakthroughs such as dynamic connectivity mapping, real-time neural decoding, and
closed-loop brain–computer systems that adaptively respond to neural states. However,
challenges persist, including issues of data integration, ethical considerations, and the
“black-box” nature of many AI systems, underscoring the need for transparent, equitable,
and interdisciplinary approaches. By synthesizing the latest breakthroughs and identifying
future opportunities, this review charts a path forward for the integration of AI and neuro-
science. From harnessing multimodal data to enabling cognitive augmentation, the fusion
of these fields is not just transforming brain science, it is reimagining human potential.
This partnership promises a future where the mysteries of the brain are unlocked, offering
unprecedented advancements in healthcare, technology, and beyond.
Keywords: artificial intelligence; neuroscience; brain–computer interfaces; neuroimaging;
neural signal processing; personalized medicine; neurological disorders; explainable AI;
cognitive augmentation; multimodal data integration
J. Clin. Med. 2025,14, 550 https://doi.org/10.3390/jcm14020550
J. Clin. Med. 2025,14, 550 2 of 45
1. Introduction
1.1. Background and Motivation
Artificial Intelligence (AI) has emerged as a cornerstone in addressing the intricate
challenges of neuroscience, a field focused on deciphering the complexities of the hu-
man brain. With approximately 86 billion neurons forming trillions of synaptic connec-
tions, the brain operates as a highly dynamic, non-linear system [
1
]. Understanding its
functions—ranging from basic reflexes to higher-order cognition—relies on processing
vast amounts of data, which span modalities such as neuroimaging, electrophysiology,
and behavioral studies. Traditional analytical tools, while effective within limited scopes,
often fall short when tasked with capturing the nuanced, multi-scale patterns embedded in
neural data. This gap has driven the integration of AI as a critical tool for neuroscientific
exploration [2].
AI technologies excel at uncovering patterns and relationships in complex, high-
dimensional datasets. Among its methodologies, deep learning (DL) has proven par-
ticularly transformative, with neural network architectures such as convolutional neu-
ral networks (CNNs) enabling the precise analysis of neuroimaging data. These tools
have been used to detect structural abnormalities linked to disorders like Alzheimer’s
disease, offering diagnostic capabilities that surpass conventional techniques [
3
]. Simi-
larly, recurrent neural networks (RNNs), designed to model temporal sequences, have
enhanced our ability to interpret electrophysiological signals, making strides in areas like
epilepsy monitoring, where predicting seizure onset can significantly improve patient
outcomes [4].
Beyond its utility as an analytical tool, AI serves as a conceptual bridge, linking
neuroscience’s understanding of biological intelligence to computational frameworks.
Artificial neural networks (ANNs), inspired by the hierarchical organization of the brain,
are modeled on principles such as Hebbian learning and synaptic plasticity. These networks,
now foundational in AI, mimic the brain’s ability to process information efficiently through
interconnected layers. Advances in neuromorphic computing, which seeks to replicate
the spiking behavior of biological neurons in hardware, highlight the evolving feedback
loop between neuroscience and AI [
5
]. This bidirectional relationship underscores how
neuroscience not only benefits from AI but also inspires its evolution [6].
The practical implications of AI’s integration into neuroscience are vast. For instance,
in brain–computer interfaces (BCIs), AI algorithms decode neural activity in real time,
enabling paralyzed individuals to control external devices with their thoughts [
7
]. Such
systems leverage cutting-edge neural decoders, translating signals from brain regions
involved in motor control into precise commands [
8
]. In drug discovery, AI has streamlined
the identification of therapeutic targets for neurological disorders by analyzing genetic,
proteomic, and clinical data. Reinforcement learning (RL) algorithms have been particularly
effective in optimizing molecular candidates, accelerating the development of treatments
for conditions such as Parkinson’s disease (PD) [9].
However, the integration of AI into neuroscience is not without challenges. The inter-
pretability of AI models remains a significant concern, particularly in fields like medicine,
where understanding the rationale behind predictions is critical [
10
]. Neuroscience data,
often characterized by noise and variability, further complicate model training and valida-
tion. Ethical considerations also loom large, especially in protecting patients’ data privacy
and ensuring equity in the development and application of AI-driven tools [11].
Despite these challenges, the convergence of AI and neuroscience heralds a transforma-
tive era. By enabling more granular analyses of neural data, AI is uncovering insights that
were previously inaccessible. Its application extends beyond research, driving innovations
in clinical care and assistive technologies. This symbiosis between AI and neuroscience
J. Clin. Med. 2025,14, 550 3 of 45
not only advances our understanding of the brain but also holds promise for improving
the lives of millions affected by neurological conditions. As these fields continue to evolve
together, their combined potential offers an unprecedented opportunity to reshape the
future of neuroscience research and its real-world applications [12,13].
1.2. Scope and Objectives of the Review
The intersection of AI and neuroscience represents a transformative frontier, merging
computational power with the complexity of brain science to address challenges once
thought insurmountable. Neuroscience, driven by vast and diverse datasets from neu-
roimaging, electrophysiology, and genomics, has increasingly turned to AI to unlock new
insights. AI, with its capacity to identify hidden patterns, model complex relationships,
and make accurate predictions, offers unprecedented opportunities to deepen our under-
standing of the brain while advancing practical applications in research and medicine.
This review seeks to provide a detailed exploration of the convergence between AI
and neuroscience, highlighting how these fields are not only advancing individually but
also propelling each other forward. It begins by clarifying foundational concepts, bridging
the gap for readers with diverse expertise. Core AI methodologies, including supervised,
unsupervised, and reinforcement learning, are contextualized within neuroscience, demon-
strating how these techniques address critical challenges, such as analyzing dynamic neural
systems or decoding the subtle patterns embedded in noisy data.
A central focus of this review is the transformative impact of AI on neuroscience ap-
plications. From enhancing neuroimaging precision to advancing neural signal processing
and predictive modeling, AI has enabled discoveries that were previously unattainable.
For example, tools powered by AI now facilitate the accurate mapping of neural connec-
tivity, the prediction of behavioral and cognitive outcomes, and the identification of early
biomarkers for neurological disorders. These advancements are not only refining research
techniques but also improving clinical interventions, enabling more personalized and
effective care.
While the potential is vast, the integration of AI into neuroscience is not without
its challenges. The variability and complexity of neuroscience data represent significant
hurdles for AI algorithms, which require clean, high-quality, and standardized datasets to
perform optimally. The “black-box” nature of many AI models poses additional concerns,
particularly in medical applications where interpretability and trust are essential. Ethical
considerations, including patient privacy, data security, and algorithmic bias, further com-
plicate the landscape. This review examines these barriers, offering a critical perspective on
how they might be addressed to ensure the responsible and effective use of AI.
Looking toward the future, this review explores emerging directions that promise
to shape the next wave of progress at the AI–neuroscience interface. These include the
development of explainable AI (XAI) models, which aim to improve transparency with-
out sacrificing performance, and the integration of diverse data modalities—combining
imaging, electrophysiological, and genetic data—to create a more holistic understanding
of the brain. The application of AI in personalized neuroscience is also discussed, high-
lighting its potential to tailor treatments to individual neural profiles, paving the way for
breakthroughs in precision medicine.
Ultimately, this review aims to inspire and inform, synthesizing recent advancements
while identifying opportunities for innovation. By providing actionable insights and
fostering interdisciplinary collaboration, it seeks to contribute to the ongoing evolution of
both AI and neuroscience, ensuring that their combined potential is harnessed to address
critical questions about the brain and its disorders. This is not merely a synthesis of past
J. Clin. Med. 2025,14, 550 4 of 45
achievements but a call to action, encouraging researchers, clinicians, and technologists to
continue pushing the boundaries of what AI and neuroscience can achieve together.
1.3. Structure of the Review
This review is organized to offer a coherent and in-depth exploration of the dynamic
intersection between AI and neuroscience, guiding readers through foundational knowl-
edge to cutting-edge applications and future possibilities. Each section is designed to build
seamlessly upon the last, creating a flowing narrative that reflects the evolving relationship
between these fields.
The journey begins with an Overview of AI, where the fundamental principles of AI are
introduced. Key methodologies, such as machine learning (ML) and neural networks, are
outlined in a way that highlights their relevance to neuroscience. By tracing the historical
development of AI and emphasizing its current capabilities, this section provides a strong
foundation for understanding its integration into neuroscience.
Following this, the Overview of Neuroscience delves into the essential concepts
that underpin the study of the brain. It explores how neural structures and processes
produce complex behaviors and cognitive functions while generating vast and diverse
datasets. These include neuroimaging outputs, electrophysiological recordings, and genetic
information, all of which present analytical challenges that AI is uniquely positioned to
address. This section sets the stage for understanding the transformative potential of AI
in neuroscience.
The next section, The Intersection of AI and Neuroscience, examines how these dis-
ciplines inspire and enhance one another. It discusses how biological principles from
neuroscience have influenced AI architectures and, in turn, how AI has become an invalu-
able tool for decoding brain activity and modeling neural systems. By presenting this
bidirectional relationship, this review illuminates the collaborative nature of progress at
this interface.
At the heart of this review is the Applications of AI in Neuroscience, which explores
real-world breakthroughs enabled by AI. This section highlights specific areas where AI
has proven instrumental, from advancing neuroimaging analysis to improving neural
signal processing and enabling brain–computer interfaces. It also examines how AI-driven
computational modeling is shedding light on complex neural behaviors, supporting both
research and clinical advancements.
To ground these discussions in practical outcomes, the Case Studies section presents
significant examples of AI’s contributions to neuroscience. These include pioneering studies
and collaborative projects that have reshaped the understanding and treatment of brain
disorders. By analyzing these achievements, this review underscores the real-world impact
of integrating AI into neuroscience.
The Challenges and Limitations section offers a critical perspective on the hurdles
that remain. It explores the technical difficulties of managing diverse and noisy datasets,
the interpretability challenges of AI models, and the ethical questions surrounding their
application. This section highlights the importance of addressing these issues to ensure the
responsible and equitable use of AI in neuroscience.
Our paper concludes with a forward-looking Future Directions and Opportunities
section. Here, emerging trends such as XAI, multimodal data integration, and personalized
neuroscience are explored. These advancements promise to push the boundaries of what
AI and neuroscience can achieve together, paving the way for a new era of innovation.
By synthesizing these elements into a cohesive structure, this review provides a
clear and engaging roadmap for understanding the evolving relationship between AI and
neuroscience. Each section contributes to a deeper appreciation of the transformative
J. Clin. Med. 2025,14, 550 5 of 45
potential at this intersection, while also identifying challenges that require collective effort
to overcome.
2. Overview of AI
AI has emerged as a transformative force across scientific disciplines, with neuro-
science standing out as one of its most promising applications. The unparalleled ability
of AI to decipher complex, multidimensional data has revolutionized the way researchers
analyze, interpret, and model brain functions [
14
]. By bridging the gap between raw data
and actionable insights, AI offers a powerful toolkit for addressing the challenges inherent
in neuroscience, a field characterized by its intricate structures and dynamic processes. This
section provides a detailed yet fluid exploration of AI’s foundational principles, historical
evolution, and specialized methodologies that have profoundly influenced neuroscience
research [15].
2.1. Definition and Core Concepts
AI can be understood as a suite of computational systems designed to replicate human
cognitive abilities such as reasoning, learning, and problem-solving. At its heart, AI relies
on algorithms that adapt and improve with experience, rather than relying solely on pre-
defined instructions. In neuroscience, AI serves as a crucial bridge, transforming massive,
intricate datasets into meaningful patterns and predictions, ultimately advancing our
understanding of the brain [16].
Two fundamental pillars of AI are ML and DL. ML focuses on teaching algorithms
to identify patterns and relationships in data. One key approach is supervised learning,
where models are trained on datasets labeled with known outcomes. In neuroscience,
supervised learning is commonly used for diagnostic tasks, such as predicting Alzheimer’s
disease (AD) progression from MRI data by identifying subtle structural changes [
17
].
On the other hand, unsupervised learning thrives in scenarios where the data lack clear
labels, uncovering hidden structures within complex datasets. For example, clustering
algorithms can reveal unique connectivity patterns across brain networks, shedding light
on differences between healthy individuals and those with neurological conditions [18].
Deep learning, a specialized subset of ML, mimics the hierarchical nature of neural
processing in the brain. ANNs, the backbone of deep learning, consist of interconnected
layers that extract progressively complex features from input data. CNNs, for instance,
are particularly effective in analyzing spatially structured data like neuroimaging scans,
identifying lesions, or segmenting brain regions [
19
]. RNNs, another type of DL model,
excel at capturing temporal dependencies, making them ideal for decoding time-series
data such as EEG recordings. Transformers, a more recent innovation, leverage attention
mechanisms to integrate diverse datasets, allowing for holistic analyses of brain function
across multiple modalities [20].
2.2. Historical Development of AI
The story of AI is one of iterative innovation shaped by theoretical breakthroughs,
technological advancements, and interdisciplinary applications. It began in the mid-20th
century, with early AI systems focusing on symbolic reasoning and rule-based logic. These
early models, such as the Logic Theorist and General Problem Solver, demonstrated the
potential of computational reasoning but were constrained by their inability to handle
real-world, unstructured data [21].
The 1980s marked a pivotal era with the introduction of artificial neural networks.
Inspired by simplified representations of biological neurons, these networks promised to
model more adaptive learning processes. However, their growth was initially stifled by
J. Clin. Med. 2025,14, 550 6 of 45
limited computational resources and insufficient data availability. The advent of backprop-
agation, a method for fine-tuning the network’s parameters, rejuvenated interest in neural
networks, setting the stage for the DL revolution [22].
In the late 2000s, a confluence of factors—including advances in computing power,
the rise of big data, and refinements in training algorithms—ushered in a golden age for AI.
CNNs became the standard for image-based tasks, transforming neuroimaging analyses by
enabling precise segmentation and anomaly detection. RNNs expanded AI’s reach into
sequential data, such as modeling neural activity over time [
23
]. Transformers, emerging in
the past decade, have redefined AI’s capabilities, offering unparalleled flexibility in process-
ing and synthesizing multimodal datasets. These advancements have firmly established AI
as a cornerstone in neuroscience, enabling breakthroughs in understanding brain function
and dysfunction [24].
These limitations were particularly evident in handling unstructured data, which lack
a predefined organization and include formats such as text, images, and time-series data.
For instance, textual data, like natural language, posed challenges due to their complex
syntax, semantics, and contextual dependencies, making tasks like translation or sentiment
analysis difficult for early systems [
25
]. Similarly, images, with their high-dimensional
pixel arrays, required sophisticated methods to extract meaningful features such as edges
or textures. Early AI models struggled to analyze such visual data effectively, hindering
applications like image recognition or medical imaging [
26
]. Time-series data, such as EEG
signals or financial trends, added another layer of complexity due to their sequential and
dynamic nature. Capturing temporal dependencies within these datasets was beyond the
capability of symbolic reasoning or rule-based systems [27].
The advent of backpropagation, a method for fine-tuning network parameters, re-
juvenated interest in artificial neural networks, setting the stage for the deep learning
revolution [
28
]. Backpropagation is considered revolutionary because it enables networks
to learn by systematically adjusting weights through gradient descent, efficiently minimiz-
ing errors between predicted and actual outputs [
29
]. This breakthrough addressed the
long-standing challenge of training multi-layered networks, unlocking their potential to
model complex, non-linear relationships in data [30].
The development of machine learning, and later deep learning, transformed the
landscape of unstructured data processing. CNNs addressed image processing challenges
by extracting spatial features, while RNNs introduced mechanisms to retain memory,
enabling the analysis of sequential data [
31
,
32
]. These innovations marked a pivotal
step forward, bridging the gap between early models and the complexities of real-world
data [33].
2.3. Types of AI Techniques Relevant to Neuroscience
AI’s success in neuroscience is rooted in its diverse methodologies, each tailored to
address specific challenges posed by the complexity of the brain and its data. These ap-
proaches have empowered researchers to decode neural signals, predict cognitive outcomes,
and simulate intricate brain networks [34].
Supervised learning is a cornerstone of AI applications in neuroscience. By training on
labeled datasets, supervised models excel in diagnostic and predictive tasks. For instance,
they are used to identify early biomarkers of neurodegenerative diseases, such as subtle
structural changes in brain scans that signal cognitive decline. These models have also been
deployed to classify neuronal activity patterns, helping researchers link brain function to
specific behaviors or cognitive states [35].
Unsupervised learning, in contrast, is the method of choice for exploratory neuro-
science. It is particularly valuable for uncovering latent patterns in neural connectivity
J. Clin. Med. 2025,14, 550 7 of 45
or simplifying the complexity of large-scale datasets, such as single-cell gene expression
profiles. By clustering similar data points or reducing dimensions, unsupervised algorithms
have revealed novel insights into brain organization and function [36].
Unsupervised learning, in contrast, is the method of choice for exploratory neuro-
science. It is particularly valuable for uncovering latent patterns in brain connectivity
or simplifying the complexity of large-scale datasets, such as single-cell gene expression
profiles [
37
]. Common algorithms include k-means clustering, which groups similar data
points to reveal subtypes of neurons or patterns in brain connectivity, and principal com-
ponent analysis (PCA), a dimensionality reduction technique often employed to identify
dominant trends in high-dimensional neuroimaging or genomic data [38].
Another widely used method is t-distributed stochastic neighbor embedding (t-SNE),
which is especially effective for visualizing high-dimensional data in two or three di-
mensions [
39
]. For example, t-SNE has been applied to single-cell transcriptomic data,
enabling researchers to distinguish neuronal subtypes and map their spatial relationships.
These algorithms play a crucial role in transforming complex, noisy datasets into inter-
pretable formats, allowing neuroscientists to uncover hidden structures and generate new
hypotheses [40].
RL offers a unique perspective, drawing inspiration from the brain’s reward systems.
This technique trains AI agents to make sequential decisions by maximizing rewards,
mirroring how organisms learn through trial and error [
41
]. Neuroscientists have used RL
to model decision-making processes, decode neural signals linked to reward pathways,
and understand how adaptive behaviors emerge from neural activity [42].
While RL has shown remarkable promise in neuroscience, its implementation is
accompanied by unique technical challenges. One primary issue is the sample inefficiency
of RL algorithms, which often require extensive interactions with an environment to learn
optimal policies [
43
]. In neuroscience, this presents significant barriers, as data acquisition
from participants or animal models is often time-intensive, ethically constrained, or limited
in scale. Using high-fidelity simulations to generate synthetic data has emerged as a
potential solution, but these models often fail to capture the full complexity of biological
systems, limiting their applicability to real-world neural processes [44].
Another challenge is the computational intensity of RL, especially in applications re-
quiring large-scale neural data, such as optimizing BCIs or simulating decision-making pro-
cesses [
45
]. Training RL models on high-dimensional datasets like fMRI or EEG recordings
can be resource-intensive, requiring advanced hardware such as GPUs or neuromorphic
platforms to manage the memory and processing demands. Strategies such as paralleliza-
tion or asynchronous learning can reduce the computational overhead, making RL more
accessible to research teams with limited resources [46].
A notable obstacle is the generalization gap, where RL models struggle to adapt
learned policies to new datasets or experimental conditions. Neural variability across
individuals or sessions often leads to inconsistent model performance, such as a BCI
trained for one individual’s neural patterns failing to generalize to another [
47
]. Promising
approaches include meta-learning, which enables models to adapt rapidly to new condi-
tions, and transfer learning, which allows pre-trained models to be fine-tuned for specific
applications [48].
The interpretability of RL algorithms also poses a critical challenge, particularly in
clinical contexts where understanding a model’s decisions is essential. This is especially
relevant in areas like drug development, where RL is used to identify candidate molecules.
Integrating XAI techniques, such as saliency maps or feature attribution, can make RL
outputs more transparent and actionable for clinicians and researchers, fostering greater
trust in these systems [49,50].
J. Clin. Med. 2025,14, 550 8 of 45
Finally, the reward design problem remains a significant limitation in neuroscience
applications. RL models rely on carefully crafted reward signals to guide learning, but
defining appropriate rewards can be complex in tasks like adaptive neural stimulations or
optimizing therapeutic interventions [
51
]. Poorly designed reward structures can lead to
unintended or suboptimal behavior. Techniques such as inverse reinforcement learning
(IRL), which derives rewards from expert demonstrations, and multi-objective RL, which
balances competing goals, are emerging as practical solutions to address this issue [52].
By addressing these challenges, RL can expand its utility in neuroscience, bridging
the gap between theoretical potential and real-world applications. Continued innovation
in simulation techniques, hardware efficiency, and model adaptability will be essential to
overcome these barriers.
CNNs are indispensable in neuroimaging, where they enhance the analysis by automat-
ing tasks like tumor detection, segmentation of brain structures, and mapping functional
connectivity [
53
]. These networks have also been adapted to analyze advanced imag-
ing modalities, such as diffusion tensor imaging, to explore microstructural properties of
the brain [54].
RNNs and their variants, such as long short-term memory (LSTM) networks, are vital
for analyzing sequential neural data [
55
]. They excel in capturing dynamic relationships
over time, enabling predictive modeling of neural activity, such as forecasting seizure
events or understanding oscillatory brain activity during cognitive tasks [56].
Despite their successes, CNNs and RNNs face significant challenges in their applica-
tion to neuroscience. One major limitation is their reliance on large, high-quality datasets to
achieve optimal performance. In neuroscience, acquiring such datasets is challenging due
to the labor-intensive nature of data collection and privacy constraints [
57
,
58
]. For instance,
labeling EEG data requires extensive manual effort by experts, while neuroimaging datasets
are often limited in availability, especially for rare neurological conditions.
Data standardization also represents a critical hurdle. Variability in EEG electrode con-
figurations, MRI scanner settings, or participant demographics introduces inconsistencies
that can hinder model generalization across studies. These discrepancies often necessitate
substantial preprocessing and augmentation, adding complexity to the workflow [59].
High-dimensional data, such as MRI scans or long EEG recordings, further complicate
computational requirements. Training these models on large-scale datasets demands
significant memory and processing power, which can limit accessibility for smaller research
teams. Additionally, overfitting is a persistent risk, particularly when models are trained
on smaller datasets, reducing their reliability for unseen data [60].
Addressing these limitations requires innovative solutions, such as leveraging transfer
learning to make use of pre-trained models and establishing standardized protocols for
data collection and preprocessing. These advancements are vital to fully realizing the
potential of CNNs and RNNs in neuroscience.
Transformers have emerged as a game changer in neuroscience, with their attention
mechanisms allowing them to analyze complex relationships across entire datasets. This
architecture has proven particularly effective in integrating multimodal data, combining
neuroimaging data, genetic profiles, and behavioral metrics to provide a more comprehen-
sive understanding of brain function and disease mechanisms [61].
Generative models, including Variational Autoencoders (VAEs) and Generative Ad-
versarial Networks (GANs), offer creative solutions for generating realistic neural activity
or synthetic datasets [
62
]. These techniques are invaluable in scenarios where real-world
data are scarce, such as for rare neurological conditions, and provide a sandbox for testing
hypotheses or training other AI systems [63].
J. Clin. Med. 2025,14, 550 9 of 45
While generative models such as GANs and VAEs have revolutionized the creation
of synthetic datasets, their use is accompanied by challenges that must be addressed for
broader applicability in neuroscience. A key concern is the accuracy and biological validity
of synthetic data [
62
]. Although these models aim to replicate the statistical properties of
real data, their output can lack subtle patterns inherent in complex neural datasets. For
instance, synthetic fMRI data generated from limited training datasets may fail to reflect
the variability and noise present in real-world data, potentially introducing biases into
downstream analyses [64].
Another significant issue is the sim-to-real gap, where models trained on synthetic
data struggle to perform well on real-world tasks due to discrepancies in data distributions.
This gap can limit the generalizability of models trained solely on synthetic datasets [
65
].
A promising solution is the development of hybrid datasets, which combine synthetic
data with smaller amounts of real data to enhance training quality while mitigating the
reliance on large, hard-to-obtain datasets. Domain adaptation techniques are also being
explored to align synthetic data more closely with real-world distributions, improving
model robustness [66].
Furthermore, synthetic data use raises concerns about validation and clinical trust. In
applications such as diagnostic tool development, models trained on synthetic data may
face scrutiny regarding their reliability and safety [
67
]. Establishing rigorous validation
pipelines, including independent benchmarking and testing against diverse real-world
datasets, is essential to ensure clinical acceptance. Collaborative efforts between AI re-
searchers and neuroscientists can help refine synthetic data pipelines and align them with
the standards required for clinical and translational applications [68].
Despite these challenges, synthetic data remain a transformative tool for neuroscience,
particularly in scenarios where real-world data are scarce or ethically constrained. Ad-
dressing these issues will unlock the full potential of generative models, allowing them
to play a pivotal role in accelerating discoveries and advancing AI-driven neuroscience
research [69].
These methodologies have transformed neuroscience, enabling researchers to navigate
the complexity of brain data with precision and efficiency. By leveraging AI’s versatility,
neuroscientists are not only unraveling the mysteries of the brain but also developing tools
and insights that promise to revolutionize research and clinical care [70].
3. Overview of Neuroscience
Neuroscience is the intricate study of the nervous system, striving to uncover the
principles that govern brain function, cognition, behavior, and disease. With its vast scope,
the field draws on diverse disciplines, including molecular biology, neuroimaging, and
behavioral science, weaving these perspectives into a cohesive effort to understand one
of the most complex systems known to science. Recent technological breakthroughs and
computational advances, particularly the integration of AI, are accelerating discoveries,
allowing researchers to tackle long-standing mysteries with newfound clarity [14,71,72].
3.1. Fundamental Concepts in Neuroscience
The nervous system is a marvel of complexity, comprising billions of neurons and
glial cells interconnected in dynamic and adaptive networks. Neurons, the brain’s primary
signaling units, communicate through electrical impulses that travel along axons and
trigger the release of neurotransmitters at synapses. These chemical signals bridge the gap
between neurons, facilitating the flow of information. Meanwhile, glial cells, traditionally
viewed as support elements, have emerged as active players, regulating synaptic activity,
maintaining neural homeostasis, and orchestrating responses to injury [73].
J. Clin. Med. 2025,14, 550 10 of 45
The brain’s adaptability, known as neuroplasticity, is one of its defining features. This
ability to reorganize and reshape itself in response to learning, environmental changes,
or damage occurs at multiple levels [
74
]. Synaptic plasticity, driven by mechanisms like
long-term potentiation (LTP) and long-term depression (LTD), strengthens or weakens
connections based on activity, forming the foundation of learning and memory.
Among the mechanisms underlying neuroplasticity, LTP and LTD are pivotal processes
that enable the strengthening or weakening of synaptic connections. These mechanisms
are studied through precise experimental techniques designed to capture synaptic changes
at molecular, cellular, and network levels [
75
]. Electrophysiological methods, such as
patch-clamp and field recordings, are commonly employed to monitor synaptic responses
during LTP and LTD induction. These approaches, often applied to brain slices or cultured
neurons, allow researchers to measure synaptic strength in real-time and to assess the
effects of specific stimulation protocols, such as high-frequency stimulation (HFS) for LTP
or low-frequency stimulation (LFS) for LTD [76].
To explore these mechanisms in living systems,
in vivo
techniques such as calcium
imaging and two-photon microscopy are increasingly utilized. These methods enable
the visualization of synaptic activity and structural remodeling, such as dendritic spine
changes, within intact neural circuits. By capturing these processes during behavior or
environmental interactions, researchers can connect LTP and LTD to their functional roles
in memory and learning [77,78].
However, studying LTP and LTD presents several challenges. Electrophysiological
recordings, while precise, are inherently invasive and often restricted to small neuronal
populations, limiting insights into broader network interactions [
79
].
In vivo
imaging,
though non-invasive, faces difficulties with signal-to-noise ratios and spatial resolution,
particularly in deeper brain structures [
80
]. Additionally, variability in experimental proto-
cols, including differences in animal models, synapse types, and stimulation parameters,
often complicates reproducibility and cross-study comparisons [81].
Advancing the study of LTP and LTD requires the standardization of protocols and
the integration of complementary approaches. Computational modeling offers a promising
avenue for bridging scales, enabling researchers to simulate plasticity mechanisms and
interpret experimental data more effectively. By addressing these challenges, researchers
can further unravel the roles of LTP and LTD in shaping adaptive neural circuits and
behavior [82–84].
Beyond the synapse, structural plasticity involves the remodeling of dendrites and
axons, as seen during recovery from brain injury or the acquisition of new skills [
85
]. Recent
advances in imaging have unveiled nanoscale transformations within synaptic structures,
shedding light on how experiences shape the brain.
While specific brain regions are specialized for particular tasks—such as the hippocam-
pus for memory encoding or the visual cortex for processing visual stimuli—these regions
seldom act in isolation. Instead, they collaborate as part of dynamic networks, coordinating
activity to perform complex functions [
86
]. Functional MRI (fMRI) studies have revealed
key networks like the default mode network (DMN), which is active during introspection
and memory retrieval, and the salience network, which guides attention to critical external
stimuli. While fMRI has significantly advanced our understanding of functional brain
networks, it has inherent limitations that must be considered. One key challenge is its
limited temporal resolution, as fMRI relies on measuring blood-oxygen-level-dependent
(BOLD) signals, which reflect metabolic changes rather than the millisecond-scale electrical
activity of neurons [
87
]. This delay makes it difficult to capture rapid, dynamic interactions
between networks, particularly during complex cognitive tasks [88].
J. Clin. Med. 2025,14, 550 11 of 45
Additionally, fMRI’s spatial resolution, although effective for identifying large-scale
networks, is less precise when examining finer details, such as the microcircuitry within spe-
cific brain regions. Artifacts near air–tissue interfaces, such as those around the orbitofrontal
cortex, can further reduce accuracy [89,90].
Another limitation lies in the correlational nature of functional connectivity anal-
yses. While these methods reveal statistical associations between regions, they cannot
establish causal relationships. Effective connectivity techniques, like dynamic causal mod-
eling (DCM), address this gap but often require complex computational modeling and
assumptions, which can vary across studies [91,92].
To overcome these challenges, researchers increasingly integrate fMRI with comple-
mentary methods like EEG and MEG. These multimodal approaches combine fMRI’s
strength in spatial resolution with the high temporal precision of electrophysiological
techniques, offering a more holistic view of network dynamics [93,94].
Interactions between brain networks, such as the DMN and the salience network, are
crucial for coordinating complex cognitive processes. These interactions are often studied
using functional connectivity analyses derived from fMRI [
95
]. Functional connectivity
examines statistical correlations between neural activities in different regions, revealing
how networks collaborate during rest or task engagement. For instance, during attention-
demanding tasks, the salience network identifies relevant external stimuli and facilitates the
transition from the internally focused DMN to task-oriented networks like the frontoparietal
network [96].
To uncover directional influences between networks, researchers employ effective
connectivity techniques, such as DCM and Granger causality. These methods help them
map how activity in one network influences another, showing, for example, how the
salience network modulates DMN activity to optimize cognitive performance [97].
Complementary insights are provided by EEG and MEG, which capture network
interactions at higher temporal resolutions. Cross-frequency coupling (CFC), where slower
oscillations influence faster rhythms, has been observed between the DMN and salience
network during tasks requiring attention shifts. This synchronization supports the rapid
reconfiguration of brain states essential for cognitive flexibility [98].
Studying network interactions comes with challenges, such as the computational
complexity of connectivity models and the need for rigorous statistical controls to mitigate
spurious findings [
99
]. Moreover, differences in imaging modalities and analytical assump-
tions can yield inconsistent results across studies. Multimodal approaches that combine
fMRI, EEG, and MEG are increasingly used to overcome these limitations, offering a more
comprehensive view of how brain networks dynamically interact to support cognition.
Understanding these networks has shifted the focus from isolated regions to the intricate
choreography of brain-wide interactions [100].
Oscillatory activity, or brain rhythms, is another cornerstone of neuroscience. These
rhythmic patterns of neural firing, categorized by frequency bands such as theta, alpha, and
gamma, synchronize activity across regions, enabling processes like attention, sensory inte-
gration, and working memory [101]. Cross-frequency coupling, where slower oscillations
regulate faster ones, has been identified as a mechanism for coordinating information flow
during cognitive tasks. Disruptions in these rhythms are increasingly linked to neurological
and psychiatric conditions, offering new avenues for therapeutic interventions [102].
3.2. Current Challenges in Neuroscience Research
Despite remarkable progress, neuroscience remains a field defined by its challenges.
One of the most significant hurdles is bridging the gap between molecular-level discoveries
and systems-level understanding. While techniques like single-cell RNA sequencing
J. Clin. Med. 2025,14, 550 12 of 45
(scRNA-seq) and optogenetics provide detailed insights into cellular functions, linking
these findings to whole-brain dynamics and behavior is a complex and ongoing endeavor.
This challenge is compounded by the multiscale nature of neural activity, which spans
nanometer-sized synaptic changes to interactions between entire brain regions [103,104].
Temporal resolution also poses a persistent challenge. fMRI offers exquisite spatial
detail but captures neural activity indirectly and with delays due to the hemodynamic
response. Conversely, techniques like EEG and MEG provide millisecond precision but lack
detailed spatial localization. Hybrid approaches, such as simultaneous EEG-fMRI record-
ings, are emerging as promising solutions, though they require advanced computational
frameworks for effective integration [105].
Another layer of complexity arises from the inherent variability of the nervous system.
Individual differences in brain anatomy, connectivity, and activity patterns are shaped by
genetics, development, and life experiences. For example, even in individuals with the
same neurological condition, the underlying neural mechanisms may differ significantly,
complicating the development of generalized models [
106
]. Personalized neuroscience,
which combines data from neuroimaging, genomics, and behavior, offers a way forward,
tailoring models to account for these variations [107].
The translation of neuroscience findings into clinical practice remains an ongoing
challenge. Despite major strides in understanding diseases like Alzheimer’s disease and
epilepsy, effective treatments remain elusive. A critical bottleneck is the lack of reliable
biomarkers that can predict disease onset, progression, or the response to therapies. Ethical
concerns surrounding emerging technologies, such as BCIs and neuromodulation tech-
niques, also present challenges. These innovations raise important questions about privacy,
autonomy, and equitable access, emphasizing the need for careful ethical considerations as
the field advances [108].
3.3. Data Types and Sources in Neuroscience
The diversity of data sources in neuroscience reflects the complexity of the brain and
the multifaceted approaches needed to study it. Each modality provides unique insights,
and their integration is increasingly essential for advancing our understanding of the
nervous system [109].
Neuroimaging has revolutionized how researchers study the brain. Structural MRI
offers detailed maps of anatomy, revealing changes associated with aging, trauma, or
disease. fMRI tracks activity-related changes in blood flow, providing insights into how
brain regions communicate during tasks or at rest. Diffusion tensor imaging (DTI) traces
white matter pathways, illuminating the structural connections that support neural commu-
nication. Positron emission tomography (PET), often paired with MRI, enables researchers
to study molecular processes, such as the buildup of amyloid plaques in AD or dopamine
dysregulation in Parkinson’s disease [110–112].
Electrophysiological techniques provide an unparalleled view of neural activity at
high temporal resolution. Non-invasive methods like EEG and MEG capture electrical and
magnetic signals generated by neuronal ensembles, making them invaluable for studying
oscillatory activity or event-related responses [
113
]. Invasive approaches, such as multi-
electrode arrays and patch-clamp recordings, allow researchers to measure the activity of
individual neurons or small networks with exceptional precision, offering a window into
neural coding during complex behaviors [114].
Genomic and molecular approaches are transforming neuroscience at the cellular
level. scRNA-seq has revealed the diversity of neuronal and glial cell types, linking gene
expression patterns to specific roles within circuits. Genome-wide association studies
(GWASs) have identified genetic variants associated with neuropsychiatric disorders, pro-
J. Clin. Med. 2025,14, 550 13 of 45
viding targets for therapeutic development. These molecular insights are increasingly
integrated with functional data, offering new perspectives on how genetic factors shape
neural systems [115,116].
Behavioral data remain vital for understanding how the brain drives real-world
actions. Advances in wearable technologies and virtual reality environments have enabled
researchers to study behaviors like navigation, decision-making, and social interaction in
more naturalistic settings. These approaches provide richer datasets, allowing researchers
to link neural activity to ecologically relevant behaviors [117,118].
Emerging technologies are redefining how data are collected and interpreted in neu-
roscience. Volumetric calcium imaging enables the real-time tracking of neural activ-
ity across large populations of neurons, while whole-brain clearing methods, such as
CLARITY, provide detailed three-dimensional maps of neural circuits [
119
,
120
]. Hybrid
modalities, like optogenetics–functional MRI, allow researchers to establish causal links be-
tween neural activation patterns and behaviors, bridging the gap between observation and
intervention [121].
The integration of these diverse datasets presents challenges but also unparalleled
opportunities. AI-powered approaches are increasingly critical for managing and analyzing
multimodal data, identifying subtle patterns, and creating predictive models that span
scales from molecules to behavior [
2
]. For example, ML algorithms are being used to com-
bine PET imaging, EEG signals, and genetic data to predict the progression of neurological
diseases, offering new avenues for personalized medicine and early intervention [122].
4. The Intersection of AI and Neuroscience
The intersection of AI and neuroscience represents a vibrant and transformative col-
laboration, one that is reshaping our understanding of both fields. Neuroscience, with
its profound insights into the workings of the brain, provides inspiration and biological
grounding for computational models, while AI offers the analytical power to tackle the com-
plexities of neural data and simulate brain processes [
13
]. This relationship has unlocked
new possibilities for exploring cognition, uncovering the roots of neurological disorders,
and even designing smarter and more adaptive technologies. Here, we explore how these
disciplines connect, influence one another, and give rise to innovations that are changing
the landscape of science and technology [16].
4.1. Theoretical Connections
AI, as we know it today, owes much of its existence to the pioneering work of neu-
roscientists who sought to decode the inner workings of the brain. The first ANNs were
inspired by the structure of biological neurons, simplifying their properties into mathemati-
cal models. Early efforts, such as the McCulloch–Pitts neuron, introduced the concept of
threshold-based activation, where a neuron “fires” when inputs cross a certain threshold.
Though rudimentary by modern standards, these ideas laid the groundwork for decades of
innovation [123].
Hebbian learning, often summarized as “neurons that fire together wire together”,
added a new dimension to AI. This principle of strengthening connections through repeated
co-activation became a cornerstone of learning algorithms, influencing everything from
unsupervised learning models to contemporary neural networks [
124
,
125
]. A more nuanced
understanding emerged with spike-timing-dependent plasticity (STDP), which emphasized
the importance of timing in shaping synaptic strength. This biologically inspired concept
paved the way for spiking neural networks (SNNs), a class of models that incorporate the
temporal dynamics of neural activity for greater computational realism [126].
J. Clin. Med. 2025,14, 550 14 of 45
Perhaps no discovery in neuroscience has influenced AI more profoundly than the
brain’s hierarchical processing of sensory information [
127
]. Hubel and Wiesel’s Nobel-
winning studies of the visual cortex revealed how neurons respond to progressively com-
plex features, from edges to patterns. This inspired the architecture of CNNs, which excel at
tasks like image recognition and neuroimaging analysis [
128
]. The brain’s influence on AI
extends even further with RL, where studies of dopamine pathways and reward prediction
errors have informed algorithms that replicate the brain’s ability to learn through feedback
and adapt to changing environments [129].
Figure 1illustrates the workflow for integrating molecular feature outputs into a
machine learning pipeline for predictive analysis. The process begins with two datasets:
a training set, which is used to build the model, and a test set, which is reserved for
evaluating its performance. Molecular features or descriptors are extracted from these
datasets using computational tools, generating structured input variables for machine
learning algorithms. These features are then processed through automated predictive algo-
rithms, involving critical steps such as attribute selection, model training, cross-validation,
and iterative model testing. These steps ensure that the model effectively learns patterns
and relationships within the data. Once trained, the model is applied to the test set to
generate predictions, which may represent properties or behaviors of the molecules. The
results, often visualized as graphs or numerical outputs, provide valuable insights for
decision-making and applications in fields such as drug discovery and materials science.
This workflow exemplifies how machine learning integrates data-driven approaches with
molecular analyses to deliver accurate and actionable predictions.
J. Clin. Med. 2025, 14, x FOR PEER REVIEW 14 of 47
inspired concept paved the way for spiking neural networks (SNNs), a class of models
that incorporate the temporal dynamics of neural activity for greater computational real-
ism [126].
Perhaps no discovery in neuroscience has influenced AI more profoundly than the
brain’s hierarchical processing of sensory information [127]. Hubel and Wiesel’s Nobel-
winning studies of the visual cortex revealed how neurons respond to progressively com-
plex features, from edges to paerns. This inspired the architecture of CNNs, which excel
at tasks like image recognition and neuroimaging analysis [128]. The brain’s influence on
AI extends even further with RL, where studies of dopamine pathways and reward pre-
diction errors have informed algorithms that replicate the brain’s ability to learn through
feedback and adapt to changing environments [129].
Figure 1 illustrates the workflow for integrating molecular feature outputs into a ma-
chine learning pipeline for predictive analysis. The process begins with two datasets: a
training set, which is used to build the model, and a test set, which is reserved for evalu-
ating its performance. Molecular features or descriptors are extracted from these datasets
using computational tools, generating structured input variables for machine learning al-
gorithms. These features are then processed through automated predictive algorithms,
involving critical steps such as aribute selection, model training, cross-validation, and
iterative model testing. These steps ensure that the model effectively learns paerns and
relationships within the data. Once trained, the model is applied to the test set to generate
predictions, which may represent properties or behaviors of the molecules. The results,
often visualized as graphs or numerical outputs, provide valuable insights for decision-
making and applications in fields such as drug discovery and materials science. This
workflow exemplifies how machine learning integrates data-driven approaches with mo-
lecular analyses to deliver accurate and actionable predictions.
Figure 1. Workflow demonstrating how molecular feature outputs are processed through machine
learning algorithms, culminating in predictions based on trained models.
4.2. AI as a Tool for Neuroscience
Figure 1. Workflow demonstrating how molecular feature outputs are processed through machine
learning algorithms, culminating in predictions based on trained models.
4.2. AI as a Tool for Neuroscience
AI has transformed neuroscience by providing tools capable of unraveling the
immense complexity of the brain. Modern neuroscience generates vast and intricate
datasets—from high-resolution brain scans to recordings of thousands of neurons firing
J. Clin. Med. 2025,14, 550 15 of 45
simultaneously. AI brings the ability to analyze, interpret, and model these data in ways
that were previously unimaginable [130].
In neuroimaging, AI has revolutionized image analysis. DL models, particularly
CNNs, have automated tasks like segmenting brain structures, detecting subtle abnor-
malities, and identifying disease biomarkers. For example, AI has been used to detect
cortical thinning in MRI scans, an early marker of Alzheimer’s disease, with a precision
that matches or exceeds expert radiologists [
131
]. Functional imaging has also benefited
from AI, with ML algorithms decoding patterns of brain activity associated with specific
cognitive states or diseases, such as schizophrenia and autism [132].
Electrophysiology, which captures the electrical activity of the brain, has seen similar
advancements. Temporal data, such as EEG recordings or neural spiking activity, pose
challenges due to their high dimensionality and complexity [
133
]. RNNs and transformers
have become indispensable for analyzing these datasets, enabling breakthroughs like
predicting seizures in epilepsy patients or decoding motor intentions for BCIs.
High-resolution EEG data offer valuable insights into neural activity, particularly
in conditions like epilepsy, but their analysis is often complicated by noise and artifacts.
Non-neural signals, such as muscle activity, eye blinks, and external interference, can
obscure critical patterns, requiring robust preprocessing to improve the data quality before
AI models are applied [134].
Preprocessing steps are critical to ensure reliable results. Techniques like bandpass
filtering isolate neural signals within key frequency ranges while removing unwanted
high-frequency noise and electrical artifacts. Independent component analysis (ICA) is
commonly used to separate neural activity from non-neural artifacts, such as eye blinks or
muscle contractions [
135
]. Wavelet-based denoising further refines the data by selectively
suppressing noise while preserving essential features of neural activity. Additionally,
normalization is employed to minimize variability across channels and sessions, ensuring
that AI models can generalize effectively across datasets [136].
AI models further enhance noise handling by leveraging advanced architectures. At-
tention mechanisms in transformers and RNNs allow these models to focus on meaningful
temporal or spatial features, effectively filtering out less relevant segments [
137
]. Convolu-
tional layers in deep learning models extract robust spatial patterns that remain consistent
despite residual noise, while data augmentation—such as adding synthetic noise during
training—enhances model robustness by teaching algorithms to differentiate between true
neural activity and artifacts [138].
These models are also adaptable to different seizure types and patient-specific patterns.
Personalized AI models trained on individual EEG data can capture unique characteristics
of a patient’s seizure activity, improving accuracy [
139
]. For broader applications, multi-
label classification approaches enable models to differentiate between seizure types, such
as focal and generalized seizures, based on distinct neural signatures. Transfer learning
allows pre-trained models to be fine-tuned for specific patients with minimal additional
data, ensuring both scalability and precision [140].
AI systems incorporating these methods have demonstrated exceptional accuracy
in seizure detection, with sensitivity rates exceeding 90% in clinical studies. This preci-
sion reduces false-positive rates and facilitates real-time applications, such as closed-loop
therapeutic systems where seizure detection triggers immediate interventions, like neural
stimulation [
141
]. By addressing both noise and adaptability, AI offers a transformative
approach to epilepsy management, combining diagnostic precision with personalized care.
AI-driven spike sorting has further streamlined the analysis of neural recordings,
allowing researchers to isolate signals from individual neurons in complex datasets [142].
J. Clin. Med. 2025,14, 550 16 of 45
AI is not just a tool for analyzing data, it is also a powerful ally in modeling the
brain itself. Graph neural networks (GNNs) have been used to simulate large-scale brain
networks, revealing how the brain’s structure influences its functional dynamics [
143
].
Computational models informed by RL have provided insights into decision-making
processes, mirroring the neural dynamics underlying behaviors like reward-seeking and
risk assessment. These models help neuroscientists test hypotheses and refine theories of
brain function, offering a deeper understanding of how neural circuits work together to
produce cognition and behavior [144].
AI’s integration into clinical neuroscience has yielded transformative advancements
in diagnostics and therapy, with several real-world implementations demonstrating
its impact.
In neuroimaging, AI has significantly enhanced the detection of early disease markers.
One notable application involves the use of CNNs to analyze MRI scans for cortical thinning,
an early indicator of Alzheimer’s disease [
64
]. A landmark study demonstrated that an
AI system achieved a diagnostic sensitivity of 94% and specificity of 92%, outperforming
many radiologists in accuracy and processing speed [145].
In patients with AD, the DMN is among the earliest functional networks to show
impairment, with disruptions closely linked to amyloid-beta (A
β
) plaque accumulation and
tau protein tangles. The DMN, which includes the posterior cingulate cortex, precuneus,
and medial prefrontal cortex, is essential for memory and self-referential thinking. Its high
baseline activity and energy demands make it particularly vulnerable to A
β
toxicity, which
interferes with synaptic signaling and disrupts network connectivity.
Research has shown that in the early stages of AD, A
β
accumulation in DMN re-
gions correlates with hyperconnectivity, as the brain attempts to compensate for emerging
damage. Over time, as tau pathology spreads along DMN connections—beginning in the
entorhinal cortex and extending to the hippocampus—this hyperconnectivity transitions to
hypoconnectivity, reflecting synaptic failure and neuronal loss [
146
]. Advanced imaging
studies have demonstrated that tau deposition in DMN nodes is associated with reduced
glucose metabolism and impaired functional connectivity, directly linking the molecular
pathology to cognitive decline [147].
AI has been pivotal in studying these complex interactions. Machine learning models
have been applied to combined PET and fMRI datasets, enabling the identification of
subtle connectivity changes associated with amyloid and tau pathology [
148
]. For instance,
AI algorithms have successfully predicted the progression of AD by quantifying how
molecular disruptions influence DMN connectivity. Unsupervised clustering techniques
have also been employed to stratify patients into subgroups based on molecular and
functional profiles, supporting personalized predictions of the disease trajectory [149].
Furthermore, AI simulations have been used to assess the potential impacts of thera-
peutic interventions targeting amyloid or tau. These models provide valuable insights into
how treatments might preserve DMN connectivity and mitigate cognitive decline, empha-
sizing AI’s role not only in diagnostics but also in guiding intervention strategies [150].
By automating the identification of subtle structural changes that are often missed dur-
ing manual review, AI enables clinicians to diagnose Alzheimer’s disease earlier, offering
opportunities for timely therapeutic interventions and improved patient outcomes [151].
Electrophysiology has also benefited from AI-driven advancements, particularly in
BCIs. One clinical trial involved patients with tetraplegia using a BCI powered by RNNs to
control robotic prosthetic limbs. The AI system translated neural activity patterns from the
motor cortex into precise motor commands, allowing participants to perform tasks such as
grasping objects and self-feeding [
152
]. Over the course of the trial, patients demonstrated
significant improvements in task completion rates and accuracy, highlighting the potential
J. Clin. Med. 2025,14, 550 17 of 45
of AI-enhanced BCIs to restore functional independence in individuals with severe motor
impairments [152].
In mental health, AI-driven digital therapeutics are redefining how psychological
support is delivered. Platforms such as Woebot and Wysa employ natural language
processing (NLP) algorithms to provide cognitive–behavioral therapy (CBT) interven-
tions via chat-based interfaces [
153
,
154
]. A randomized controlled trial involving over
1000 participants found that users of these platforms experienced a 25% reduction in anxiety
symptoms and a 30% improvement in mood scores over eight weeks, outcomes comparable
to traditional face-to-face therapy for patients with mild to moderate cases [
155
]. These
systems address the growing demand for mental health services, offering scalable and
accessible solutions for underserved populations.
AI has also accelerated drug discovery in neuroscience, particularly for neurode-
generative disorders. By analyzing vast molecular datasets and predicting drug–target
interactions, AI systems can identify promising therapeutic candidates with unprecedented
speed. In one case, an ML platform screened billions of compounds and identified a can-
didate drug for Alzheimer’s disease in just six months—a process that would typically
take several years using conventional methods. This drug is now undergoing preclinical
trials, demonstrating AI’s ability to expedite the development pipeline and address critical
therapeutic gaps in neurology [156].
These case studies exemplify how AI is bridging the gap between research and clinical
practice, improving patient outcomes across diverse medical domains. However, chal-
lenges remain, including ensuring the generalizability of AI models across populations,
integrating these tools into existing clinical workflows, and navigating regulatory approval
processes. Addressing these issues will be key to unlocking AI’s full potential in clinical
neuroscience [157].
Molecular neuroscience has also embraced AI, particularly in the analysis of scRNA-
seq data. ML algorithms have identified previously unrecognized neuronal subtypes and
illuminated their roles within complex circuits. By linking molecular diversity to functional
outcomes, these tools are reshaping our understanding of how genetic factors influence
brain development and dysfunction [158,159].
Despite the transformative potential of ML in analyzing single-cell data, several chal-
lenges constrain its utility. A primary issue is the high dimensionality of scRNA-seq
datasets, which often include thousands of gene expression measurements per cell [
160
].
This complexity can lead to overfitting, where models capture noise instead of meaningful
patterns, particularly in smaller datasets or when rare cell subtypes are underrepresented.
While dimensionality reduction methods, such as PCA and uniform manifold approxima-
tion and projection (UMAP), are widely used, they can oversimplify the data, potentially
masking subtle biological variations [161].
Another challenge lies in the variability between datasets. Differences in experimental
protocols, sequencing platforms, and sample preparation can introduce batch effects that
obscure true biological signals. For instance, cells with similar expression profiles may
cluster separately if derived from different experiments, complicating cross-study compar-
isons [
162
]. Methods like Harmony and ComBat have been developed to correct for these
effects, but they require significant computational resources and careful optimization to
avoid unintended distortions of the data [163].
The interpretability of ML models is also a major limitation. Advanced models like
neural networks and ensemble methods often function as black boxes, making it difficult to
discern the biological features driving predictions. This lack of transparency hinders the
identification of novel pathways or cell subtypes. Efforts to incorporate XAI tools, such as
J. Clin. Med. 2025,14, 550 18 of 45
integrated gradients and SHAP (Shapley Additive Explanations), are underway but have
yet to see widespread application in single-cell research [164].
Furthermore, the scarcity of labeled datasets poses a significant barrier to training
robust supervised ML models. Annotating single-cell data is labor-intensive and typically
requires expert validation, which is particularly challenging for rare or poorly characterized
cell populations. Consequently, many analyses rely on unsupervised or semi-supervised
approaches, which may struggle to resolve complex or overlapping cell states without
sufficient prior knowledge [165].
Finally, the computational demands of single-cell data analysis present practical
challenges. Clustering, trajectory inference, and cell–cell interaction predictions often
involve resource-intensive algorithms, making large-scale analyses inaccessible to research
teams without a high-performance computing infrastructure. Efforts to optimize these
workflows for scalability and efficiency remain an active area of research.
Addressing these challenges will require innovative solutions, such as standardized
preprocessing pipelines, improved XAI frameworks, and collaborative initiatives for data
sharing and model validation. By overcoming these limitations, ML has the potential
to further unlock the complexities of single-cell data and enhance our understanding of
molecular neuroscience.
4.3. Neuroscience-Inspired AI Innovations
While AI has empowered neuroscience, the influence flows both ways. Many of
AI’s most significant breakthroughs have drawn inspiration directly from the brain. This
interplay has led to the development of systems that are not only more powerful but also
more efficient, adaptive, and interpretable [166].
SNNs are a prime example of how neuroscience inspires AI. Unlike traditional ANNs,
which process information as continuous activations, SNNs replicate the brain’s discrete
spiking behavior. This temporal fidelity makes them highly efficient, particularly when de-
ployed on neuromorphic hardware that mimics the architecture of biological neurons [
167
].
SNNs are being explored for applications ranging from real-time robotics to energy-efficient
computing, areas where the brain’s resourcefulness is an invaluable model [168].
The brain’s ability to integrate information across multiple senses has inspired multi-
modal AI systems. These models combine data streams, such as visual, auditory, and textual
inputs, to generate richer and more contextually aware outputs. In autonomous vehicles,
for instance, multimodal AI mimics the brain’s ability to integrate sensory information,
enabling safer and more reliable navigation [169].
RL continues to evolve under the influence of neuroscience. Insights into how animals
balance exploration and exploitation have informed algorithms that train AI agents to
optimize behavior in dynamic environments. For example, deep RL models have been
used to develop agents capable of solving complex problems, such as playing advanced
strategy games or managing logistics in supply chains [170,171].
Another area of growth is continual learning, inspired by the brain’s capacity to re-
tain and integrate knowledge across experiences. Unlike traditional AI models, which
often suffer from catastrophic forgetting, neuroscience-informed algorithms incorpo-
rate mechanisms like memory consolidation and synaptic plasticity to enable incre-
mental learning [
172
]. These systems are critical for applications requiring adaptabil-
ity, such as personalized healthcare or autonomous systems that operate in changing
environments [173].
Finally, XAI, a field aimed at making ML models transparent and interpretable, has
drawn from cognitive neuroscience. By mimicking the brain’s mechanisms for attention
and decision-making, these models prioritize relevant features and provide rationales for
J. Clin. Med. 2025,14, 550 19 of 45
their outputs [
174
]. This interpretability is essential in applications like medical diagnostics,
where trust and accountability are paramount [175].
The ongoing dialogue between AI and neuroscience is driving innovations that benefit
both disciplines. Neuroscience inspires more flexible, efficient, and biologically plausible
AI models, while AI accelerates discoveries in brain science, offering tools to analyze
data, test hypotheses, and model complex processes. Together, these fields are pushing
the boundaries of what is possible, laying the groundwork for new breakthroughs in
understanding intelligence—both natural and artificial [14].
5. Applications of AI in Neuroscience
AI has revolutionized neuroscience, catalyzing groundbreaking advancements in
research, diagnosis, and treatment. By enabling more precise data analysis, uncovering
novel insights, and offering innovative solutions, AI is redefining how we understand and
address the complexities of the brain [
176
,
177
]. This section integrates the novel and highly
relevant contributions of AI to neuroimaging, neural signal processing, brain–computer
interfaces, computational modeling, drug discovery, and cognitive–behavioral studies.
These applications are enriched with perspectives from high-impact studies to showcase
the transformative potential of AI in neuroscience.
5.1. Neuroimaging Analysis: Beyond Conventional Boundaries
AI has redefined neuroimaging by enabling the deep integration of multimodal data,
automated segmentation, and early detection of subtle biomarkers. The convergence of
structural and functional data through AI is shedding light on disease mechanisms and
improving diagnostics.
Advancing Multimodal Imaging Integration.
AI’s ability to merge diverse imaging modalities is addressing a long-standing chal-
lenge in neuroscience: linking structural, functional, and molecular brain data. A highly
cited application demonstrated how DL models integrated structural MRI with fMRI and
PET to predict cognitive decline in preclinical AD models [
178
]. By correlating amyloid
deposition PET with fMRI, these models offered unparalleled predictive accuracy, enabling
earlier interventions [1].
Dynamic Functional Connectivity Analysis.
fMRI has benefited significantly from AI-powered GNNs, which map dynamic brain
connectivity patterns over time. These networks have revealed activity disruptions in
patients with conditions like schizophrenia, where altered connectivity in the default
mode and salience networks impairs executive function [
179
]. Novel temporal analysis
techniques, such as RNNs, have captured how functional interactions evolve during
memory encoding and retrieval tasks, offering insights into conditions like mild cognitive
impairment [
180
]. Table 1summarizes key studies demonstrating AI’s pivotal role in
advancing neuroimaging technologies and methodologies.
Table 1. AI’s contributions to neuroimaging.
Reference Study Objective AI Methodology Neuroimaging
Modality
Condition
Studied
Impact on Research or
Practice
Liang et al.
[181]
Predict fMRI
responses to natural
stimuli
Deep Learning
(DNN) fMRI
Visual perception
Enhanced decoding of
visual cortical activity
for brain–
machine interfaces
J. Clin. Med. 2025,14, 550 20 of 45
Table 1. Cont.
Reference Study Objective AI Methodology Neuroimaging
Modality
Condition
Studied
Impact on Research or
Practice
Cui et al. [182]
AI-enhanced
multimodal
imaging fusion
GANs,
Transformers MRI, PET Alzheimer’s
disease
Improved the early
diagnosis by integrating
structural and
molecular biomarkers
Luo et al. [183]Multimodal imaging
for ASD
Variational
Autoencoders MRI, fMRI
Autism spectrum
disorder (ASD)
Linked functional
connectivity disruptions
to ASD-
related behaviors
Khosla et al.
[184]
Mood prediction
from
neuroimaging data
Graph Neural
Networks (GNNs)
Intracranial EEG,
fMRI Mood disorders
Enabled predictions of
mood variations from
neural activity patterns
Saidi et al.
[185]
Cross-modality
Alzheimer’s
disease detection
Convolutional
Neural Networks MRI, PET Alzheimer’s
disease
Demonstrated >90%
accuracy in early-stage
diagnosis using
AI-based
feature extraction
Gu et al. [186]
Functional
connectivity
mapping for
schizophrenia
Dynamic Causal
Models fMRI Schizophrenia
Identified disrupted
brain network
interactions associated
with cognitive deficits
Nakack et al.
[187]
Epileptic focus
localization
Support Vector
Machines (SVMs) PET Epilepsy
Improved the
identification of seizure
origins through
molecular imaging data
Chen et al.
[188]
Amyloid plaque
detection with AI
GAN-enhanced
PET reconstruction
PET Alzheimer’s
disease
Enhanced resolution
and sensitivity for
detecting early amyloid
deposition
Ranjbarzadeh
et al. [189]
Machine learning for
brain tumor
segmentation
U-Net Architecture
MRI Brain tumors
Automated
segmentation with >95%
precision, facilitating
surgical planning
Hu et al. [17]
Integrative imaging
for
neuroinflammation
Semi-Supervised
Learning MRI, PET Multiple
sclerosis
Identified
inflammation-related
imaging markers, aiding
in treatment planning
5.2. Neural Signal Processing: Unlocking Temporal Complexity
AI’s impact on neural signal processing has been transformative, particularly in its
ability to decode intricate temporal dynamics and uncover new insights into brain function.
Cutting-edge approaches like transformers are redefining how we analyze continuous
neural signals.
Decoding Oscillatory Dynamics with Transformers.
Transformers, originally developed for natural language processing, are revolutioniz-
ing the analysis of EEG and MEG data [
190
]. These models excel at capturing long-range
dependencies and interactions in neural oscillations, such as theta–gamma coupling, which
plays a crucial role in working memory and decision-making [
191
]. Recent studies have
demonstrated how transformers outperform conventional models in detecting cognitive fa-
tigue from EEG signals, providing a foundation for applications in performance monitoring
and cognitive enhancement [192].
J. Clin. Med. 2025,14, 550 21 of 45
Seizure Prediction and Real-Time Applications.
AI models trained on continuous EEG streams have achieved remarkable accuracy
in predicting epileptic seizures, a feat with profound clinical implications [
193
]. ML al-
gorithms can identify preictal patterns up to 30 min before seizure onset, giving patients
and caregivers critical time to prepare. Wearable devices powered by these AI models are
enabling real-time seizure monitoring, transforming epilepsy management and improving
patient safety [
139
]. The table below (Table 2) highlights significant studies showcasing
AI’s impact on processing and analyzing neural signals.
Table 2. AI’s applications in neural signal processing.
Reference Study Objective AI Technique Signal Type Condition
Studied
Key Results or
Contributions
Kumar et al. [194]Deep learning for
BMI signal decoding
Convolutional
Neural
Networks
EEG Motor imagery
Improved feature
extraction for motor
imagery, enhancing BMI
control accuracy
Abduljaleel et al.
[195]
Seizure detection
using deep learning CNN, RNN EEG Epilepsy
Achieved >90% seizure
prediction accuracy,
aiding in real-
time management
Matar et al. [196]
Neural oscillation
analysis for
memory studies
Transformers EEG, MEG Memory and
cognition
Identified oscillatory
patterns linked to
memory encoding
and retrieval
Alessandrini et al.
[197]
Neural correlates of
Alzheimer’s disease
via EEG
Semi-
Supervised
Learning
EEG Alzheimer’s
disease
Enhanced early
detection through neural
oscillation biomarkers
Azar et al. [198]
Real-time EEG
analysis for
emotion decoding
LSTMs EEG Anxiety,
depression
Enabled real-time
emotion monitoring
for mental
health applications
Norman et al. [199]
High-speed
decoding of
movement
intentions
Reinforcement
Learning
Electrocorticography
Paralysis
Improved accuracy in
motor intention
decoding for
assistive BCIs
Jirsa et al. [200]
Cross-frequency
coupling analysis of
meditation states
Variational
Autoencoders EEG Meditation
Identified theta–gamma
dynamics associated
with deep
meditative states
Assali et al. [201]
Preictal state
detection
for epilepsy
GANs, CNN EEG Epilepsy
Improved sensitivity to
detect seizure
precursors, aiding
preventative
interventions
Tuncer et al. [192]Cognitive fatigue
detection from EEG
Support Vector
Machines
(SVMs)
EEG Cognitive load
Developed real-time
fatigue monitoring
systems for
high-stake tasks
5.3. Brain–Computer Interfaces: Toward Seamless Integration
BCIs are leveraging AI to create more adaptive and intuitive systems for restoring
communication, mobility, and sensory feedback. These advancements are moving BCIs
from research labs to real-world applications [202].
AI-Driven Neural Decoding.
J. Clin. Med. 2025,14, 550 22 of 45
AI has elevated the performance of BCIs by improving the accuracy of neural signal
interpretation. DL models, such as CNNs, have enabled the precise control of robotic
limbs, allowing users to perform complex movements like grasping objects or navigating
environments. RL algorithms further enhance BCIs by adapting to individual neural
patterns over time, reducing errors and improving user experience [8,203].
Integrating Sensory Feedback for Natural Interaction.
A major breakthrough in BCI development is the integration of sensory feedback
systems powered by AI [
204
]. By interpreting neural activity, these systems provide users
with real-time tactile sensations, allowing for more intuitive control of prosthetic devices.
High-impact studies have shown that users equipped with AI-driven sensory feedback
experience exhibited improved dexterity and task performance, marking a significant
step toward seamless brain–machine integration [
205
]. Table 3highlights key advance-
ments in AI applications for BCIs, emphasizing their transformative potential in healthcare
and beyond.
Table 3. Advances in AI for brain–computer interfaces.
Reference Study Objective AI Technique BCI Application Condition
Studied Impact
Barnova et al. [206]
AI-driven BCIs for
movement
restoration
Reinforcement
Learning Movement control Quadriplegia
Achieved seamless
control of robotic arms
for paralyzed users
Wei et al. [207]High-accuracy
SSVEP-based BCI
Canonical
Correlation
Analysis
Communication
aids General
Improved usability of
visual BCIs for
hands-free
communication
Schiffer et al. [208]
Adaptive feedback
for
neurostimulation
Reinforcement
Learning
Neurorehabilitation
Stroke
Accelerated recovery
through real-time
neuroadaptive
feedback
Palumbo et al. [
209
]
Non-invasive BCIs
for wheelchair
navigation
Deep Learning
Models
Mobility
enhancement Paralysis
Enabled
thought-driven
wheelchair navigation
with high accuracy
Naik et al. [210]Speech decoding
via invasive BCIs CNNs, GANs Communication
aids ALS
Restored
communication
abilities for
locked-in patients
Lin et al., 2023
Neuroplasticity-
enhanced BCI
adaptability
Neuromorphic
Computing
Long-term
adaptability General
Improved long-term
performance by
incorporating
neuroplasticity
principles
George et al. [211]
Gaze-controlled
BCIs for
environmental
navigation
Transformers Accessibility tools Paralysis
Enhanced gaze-driven
BCIs, enabling
navigation and
environmental
interaction
Maye et al. [212]
Multimodal BCI
systems for
sensory
augmentation
Multimodal
Learning Models
Sensory restoration
Amputees
Integrated tactile and
auditory feedback for
advanced prosthetics
J. Clin. Med. 2025,14, 550 23 of 45
Table 3. Cont.
Reference Study Objective AI Technique BCI Application Condition
Studied Impact
Torres et al. [213]Emotion-
monitoring BCIs SVMs Mental health
tracking
Depression,
anxiety
Personalized mental
health support
through real-time
neural monitoring
Moreno-Calderón
et al. [214]
Gaming with
thought-
driven BCIs
GANs Entertainment
applications General
Delivered immersive
experiences controlled
entirely by
neural activity
5.4. Computational Modeling: Simulating Brain Dynamics
AI-powered computational models are enabling the simulation of complex neural
systems and offering insights into brain function, cognition, and disease progression. These
models are instrumental in testing hypotheses and exploring new treatment avenues [
215
].
Simulating Cortical Oscillations.
SNNs, designed to replicate the brain’s time-dependent firing patterns, have advanced
our understanding of neural oscillations. SNNs have been used to model inhibitory–
excitatory balance disruptions, which play a critical role in conditions like autism spectrum
disorders and epilepsy [
101
]. These simulations have revealed how altered synchroniza-
tion in cortical circuits contributes to sensory processing deficits and cognitive impair-
ments [216].
Predicting Neurological Disease Trajectories.
AI models integrating genetic, imaging, and clinical data are transforming our ability
to predict the progression of neurological diseases. For instance, DL algorithms have simu-
lated tau pathology spread in Alzheimer’s disease, identifying potential intervention points
that could delay cognitive decline [
217
]. Similarly, predictive models of PD have linked
dopamine depletion patterns to motor symptoms, guiding more personalized treatment
strategies [218].
5.5. Drug Discovery and Development: Accelerating Therapeutic Innovation
AI is revolutionizing drug discovery by optimizing target identification, streamlining
compound design, and improving predictions of therapeutic efficacy and safety. These
advancements are particularly impactful in addressing the challenges of neurological drug
development [219].
Identifying Novel Therapeutic Targets.
ML models trained on proteomic and transcriptomic data have identified key molecu-
lar pathways involved in synaptic dysfunction and neurodegeneration. AI has revealed
regulators of amyloid beta aggregation in patients with AD and synaptic pruning in pa-
tients with schizophrenia, offering promising targets for early-stage interventions. These
discoveries are enabling a shift toward precision medicine, where therapies are tailored to
specific molecular profiles [220,221].
Designing Safer and More Effective Drugs.
GANs and RL algorithms are driving the development of novel compounds with
optimized pharmacokinetic properties. A high-impact application demonstrated how
AI-designed antiepileptic drugs exhibited reduced off-target effects while maintaining
efficacy, accelerating their progression through preclinical trials [
222
]. These models are
transforming the speed and efficiency of drug pipelines, particularly in conditions with
complex pathophysiologies like epilepsy and multiple sclerosis [223].
J. Clin. Med. 2025,14, 550 24 of 45
5.6. Cognitive and Behavioral Analyses: Mapping Brain–Behavior Links
AI is advancing the study of cognition and behavior by uncovering connections
between neural activity, adaptive processes, and mental states. These tools are shedding
new light on decision-making, learning, and mental health [224].
Decoding Decision-Making Processes.
RL models have been used to explore how individuals adapt to changing environ-
ments, providing insights into reward-processing mechanisms. These studies have revealed
how altered dopamine signaling affects decision-making in patients with conditions like
addiction and depression. AI-driven analyses are informing the development of therapies
aimed at restoring cognitive flexibility and improving executive function [225].
Personalizing Mental Health Care.
AI is transforming mental health diagnostics by integrating neuroimaging, behavioral,
and physiological data. ML models have achieved high accuracy in predicting treatment
responses, identifying patients likely to benefit from antidepressants or CBT [
226
]. These
personalized approaches are reducing trial-and-error treatment strategies and improving
patient outcomes. Furthermore, AI tools are being applied to detect early signs of conditions
like PTSD, offering opportunities for timely interventions and preventive care [227].
6. Early Detection and Intervention Strategies Using AI: Case Studies
The integration of AI into early detection and intervention strategies has revolu-
tionized neuroscience and clinical practice. By analyzing complex datasets, uncover-
ing novel biomarkers, and enabling real-time decision-making, AI has opened new av-
enues for timely and precise healthcare [
228
]. This section explores key case studies,
each illustrating how AI is transforming the diagnosis and management of neurologi-
cal and psychiatric disorders, while maintaining a fluid narrative to highlight its unique
contributions [229].
6.1. Early Detection of Alzheimer’s Disease: Multimodal Integration and Biomarker Discovery
AD presents significant diagnostic challenges, as its early symptoms often remain
subtle and difficult to distinguish from normal aging. AI has addressed these challenges
by integrating diverse data sources to uncover patterns that predict the onset of AD long
before clinical symptoms appear [230].
In one groundbreaking application, AI models combined PET imaging, MRI scans,
and cerebrospinal fluid (CSF) biomarkers to predict which individuals were at high risk of
developing AD. By analyzing amyloid and tau deposition alongside structural connectivity
disruptions, these models achieved predictive accuracies exceeding 90%, outperforming tra-
ditional diagnostic tools. Moreover, they provided novel insights into how early molecular
changes interact with functional connectivity impairments in the brain’s DMN [231].
These advancements are reshaping the clinical landscape, enabling more precise
patient stratification for clinical trials and facilitating earlier interventions. By focusing on
those at the highest risk, AI-driven diagnostics are helping to delay or even prevent disease
progression, making profound impacts on both individuals and healthcare systems [232].
6.2. Predictive Models for Seizure Management in Patients with Epilepsy
Epilepsy management has long been reactive, relying on a retrospective analysis of
seizure patterns and limited by the unpredictability of episodes. AI has transformed this
field by enabling real-time seizure prediction, providing patients with greater autonomy
and safety [233].
One cutting-edge system utilized high-resolution EEG data and advanced neural
networks to detect preictal patterns—subtle shifts in brainwave activity that precede
J. Clin. Med. 2025,14, 550 25 of 45
seizures. By combining temporal and spatial features of brain activity, the model achieved
seizure prediction accuracies of up to 90%, offering critical early warnings up to 30 min
before onset. These capabilities were integrated into wearable devices, delivering real-time
alerts to patients via smartphone applications [234].
This innovation has reduced seizure-related injuries and improved the quality of
life of patients, enabling them to take proactive measures such as administering fast-
acting medications or avoiding potentially dangerous activities. AI-powered solutions are
fundamentally changing how epilepsy is monitored and managed, providing a new level
of control to patients [235].
6.3. Autism Spectrum Disorder: Early Detection via Neuroimaging and Behavioral Analyses
Autism spectrum disorder (ASD) is often diagnosed late, missing critical developmen-
tal windows when interventions are most effective. AI has introduced novel methods for
earlier detection by combining neuroimaging and behavioral data to identify early markers
of ASD [236].
In a prominent study, ML algorithms were applied to fMRI data from infants with
a high genetic risk of ASD. The model identified atypical connectivity patterns in brain
regions associated with social communication, such as the temporal–parietal junction and
prefrontal cortex. When these findings were combined with behavioral assessments, the
AI system achieved diagnostic accuracy rates of 85% in children under two years old, far
earlier than traditional diagnostic approaches [237].
By enabling an earlier diagnosis, these AI-driven tools allow clinicians to implement
targeted interventions during critical periods of neural and social development. Person-
alized therapies focusing on language and social skills have been shown to significantly
improve long-term outcomes for children with ASD [238].
6.4. AI-Driven Mental Health Diagnostics and Personalized Interventions
Mental health care is increasingly turning to AI for solutions to longstanding chal-
lenges, including delayed diagnosis and trial-and-error treatment strategies. By predicting
treatment outcomes and personalizing interventions, AI is improving care for conditions
such as depression and anxiety [239].
A notable study demonstrated how graph-based neural networks could analyze
functional connectivity patterns in individuals with major depressive disorder [
240
]. By
integrating these neural data with clinical histories and genetic markers, the model accu-
rately predicted responses to selective serotonin reuptake inhibitors (SSRIs) with an 80%
success rate. This capability has significantly reduced the reliance on guesswork, enabling
clinicians to tailor treatments more effectively [241].
Beyond pharmacology, AI-powered platforms for CBT are revolutionizing mental
health care. These systems adapt interventions based on user engagement and progress,
delivering personalized support in real time. By increasing accessibility and tailoring
care to individual needs, AI is breaking barriers in mental health care, particularly for
underserved populations [242].
6.5. Stroke Care: Enhancing a Rapid Diagnosis and Treatment Precision
In stroke care, every minute saved in diagnosis and treatment can dramatically im-
prove patient outcomes. AI has proven invaluable in this context, enabling faster, more
accurate diagnoses and optimizing therapeutic decisions [243].
A pioneering AI model designed for stroke diagnostics analyzed CT and MRI scans in
near real time, distinguishing between ischemic and hemorrhagic strokes with an accuracy
of 95%. The system also quantified perfusion deficits and predicted tissue viability, ensuring
J. Clin. Med. 2025,14, 550 26 of 45
that patients received appropriate treatments, such as thrombolysis, within the critical
therapeutic window [244].
Hospitals integrating AI into their stroke care protocols have reduced door-to-
treatment times by 25%, improving survival rates and minimizing long-term disability.
These advancements underscore the transformative role of AI in emergency medicine,
where speed and precision are paramount [245].
6.6. Parkinson’s Disease: Early Detection of Motor and Non-Motor Symptoms
PD is notoriously difficult to diagnose in its early stages, as its symptoms often overlap
with those of other conditions. AI has addressed this gap by identifying subtle motor and
non-motor signs that precede a clinical diagnosis [246].
A robust AI model trained on gait analysis, handwriting samples, and voice recordings
achieved diagnostic accuracies exceeding 90%, detecting early motor impairments such as
tremors and bradykinesia. By incorporating non-motor symptoms, such as sleep distur-
bances and a reduced sense of smell, the model provided a holistic diagnostic framework
that traditional methods lack [247,248].
These advancements are enabling the earlier initiation of neuroprotective treatments,
which can slow disease progression and improve quality of life. Additionally, AI models
predicting disease trajectories are guiding clinicians in tailoring treatments to individual
needs, making PD care more precise and effective [249].
6.7. Rare Neurological Disorders: AI for Diagnostic Support
Rare neurological disorders often go undiagnosed due to their complexity and the lack
of widespread expertise. AI is bridging this gap by uncovering patterns in large datasets
that point to rare conditions, offering diagnostic support where traditional methods fall
short [250].
In a striking example, an AI system developed for diagnosing Wilson’s disease—a rare
disorder of copper metabolism—analyzed genetic data, biochemical markers, and imaging
findings to achieve a diagnostic accuracy of 92%. This tool significantly reduced diagnostic
delays, which can span years, allowing for earlier treatment initiation and better patient
outcomes [251].
AI frameworks like this are being applied to other rare neurological conditions, en-
suring that even in resource-limited settings, patients benefit from early and accurate
diagnoses. These advancements are democratizing access to specialized care, addressing a
critical need in global health [252].
7. Challenges and Limitations
While AI is reshaping neuroscience and healthcare, its integration comes with a host
of challenges that demand thoughtful solutions. From data quality and interpretability
to ethical concerns and technical constraints, these obstacles must be addressed to fully
realize AI’s transformative potential [2].
7.1. Data Quality and Accessibility
AI relies on high-quality data to function effectively, but in neuroscience, the quality
and accessibility of data remain significant barriers [
253
]. Neuroimaging scans, electro-
physiological recordings, and clinical datasets are often noisy, inconsistent, and influenced
by varying acquisition protocols. These discrepancies can introduce bias and reduce the
reliability of AI models, especially when applied across diverse populations [254].
Accessibility to large, representative datasets also presents a challenge. Many datasets
are siloed within institutions or subject to strict privacy regulations, limiting their availabil-
J. Clin. Med. 2025,14, 550 27 of 45
ity for collaborative research. In resource-constrained regions, where advanced neuroimag-
ing and data collection infrastructure are scarce, this issue is even more pronounced [
255
].
Efforts like the Human Connectome Project and OpenNeuro are starting to bridge
these gaps by providing open-access, standardized datasets, but much work remains [
256
].
Federated learning—a technique where AI models are trained across decentralized datasets
without transferring sensitive data—offers a promising solution, enabling wider collabora-
tion while respecting data privacy [257].
7.2. Interpretability of AI Models
AI’s strength lies in its ability to uncover complex patterns, but its “black-box” nature
often leaves users questioning how it reaches its conclusions. This lack of transparency is
particularly problematic in neuroscience and healthcare, where clinical decisions based on
AI outputs can have life-altering consequences [
258
]. For instance, while an AI model may
predict the onset of AD with high accuracy, its inability to explain the reasoning behind the
prediction can erode trust among clinicians and patients [259].
To address this, XAI is emerging as a key focus. Techniques like attention mechanisms,
saliency maps, and interpretable feature attributions are helping to make AI systems more
transparent [
260
]. These approaches allow researchers and clinicians to better understand
how AI models weigh different data inputs, ensuring that their outputs are not only
accurate but also actionable and trustworthy [261].
7.3. Ethical and Privacy Concerns
The use of AI in neuroscience often involves handling sensitive data, such as neu-
roimaging scans, genetic profiles, and behavioral assessments. This raises pressing ethical
questions about consent, data security, and potential misuse. Patients may be hesitant
to share such intimate information, particularly if they fear breaches or misuse of their
data [262].
Bias in AI models is another critical issue. If training datasets are not representative of
diverse populations, AI systems can inadvertently perpetuate healthcare disparities, per-
forming well for some groups while failing others. For example, models trained primarily
on data from Western populations may struggle to generalize to patients from other regions
or cultural backgrounds [263].
Addressing these concerns requires a robust ethical framework. Techniques like
differential privacy, which protects individual data points within a dataset, and federated
learning can enhance data security. Meanwhile, efforts to diversify datasets are essential to
building AI systems that are equitable and effective across populations [264].
7.4. Technical and Computational Constraints
The sophisticated algorithms that power AI require significant computational re-
sources, creating barriers for their adoption in resource-limited settings. Dl models, in
particular, demand high-performance GPUs and extensive memory, which can be pro-
hibitive for smaller clinics or research institutions. Additionally, training these models is
both time-consuming and costly, limiting their scalability [265,266].
Neuroscience datasets, such as fMRI scans or EEG recordings, add another layer
of complexity. These high-dimensional datasets require advanced preprocessing and
optimization to avoid issues like overfitting, where models perform well on training data
but fail to generalize to new cases [58].
Advances in cloud computing and edge computing are beginning to alleviate some
of these challenges. Cloud platforms enable centralized processing, while edge comput-
ing brings AI capabilities directly to devices, reducing latency and resource demands.
J. Clin. Med. 2025,14, 550 28 of 45
Lightweight AI models and transfer learning techniques are also making it easier to deploy
effective systems in environments with limited computational power [267].
7.5. Integration into Clinical Practice
Despite AI’s potential, its integration into clinical workflows remains a significant
hurdle. Many clinicians are hesitant to adopt AI tools, often due to concerns about reliability,
interpretability, and disruption of established practices [
268
]. Additionally, regulatory
frameworks for approving AI in healthcare are still evolving, creating uncertainty around
compliance and accountability [269].
In practice, AI’s effectiveness often depends on its ability to complement, rather
than replace, human expertise. For instance, while an AI system may identify subtle
abnormalities in a neuroimaging scan, its insights must align with a clinician’s judgment
to ensure an accurate diagnosis and treatment. Achieving this balance requires tools
that are not only accurate but also user-friendly and seamlessly integrated into existing
workflows [270].
Training programs for healthcare professionals are helping to bridge the gap, famil-
iarizing clinicians with AI systems and their potential benefits. Regulatory agencies are
also working to establish clear guidelines, ensuring that AI tools meet rigorous safety and
efficacy standards before they reach clinical settings [271].
8. Future Directions and Opportunities
The future of AI in neuroscience holds extraordinary promise, with potential break-
throughs poised to revolutionize how we understand, diagnose, and treat brain disorders.
By building on current advancements and addressing key challenges, AI will open un-
charted frontiers in brain science [272].
8.1. Advances in AI Technologies
The next wave of AI innovation will feature advanced algorithms and architectures
designed specifically for the complexities of neuroscience.
Explainable AI and Interactive Models.
A crucial future direction is the development of AI systems that not only generate
accurate predictions but also interactively explain their reasoning [
268
]. For example,
f (XAI) models could visually map neural activity patterns that lead to specific diagnoses,
such as early Alzheimer’s, while quantifying their confidence in those predictions. These
capabilities will enhance trust and adoption in clinical settings, bridging the gap between
machine insights and human decision-making [273].
Revolutionizing Neuroscience with Generative AI.
Generative AI models, including GANs and transformers, are expected to play a
pivotal role in neuroscience. These models could simulate realistic neural data, helping
researchers test hypotheses and design experiments without relying on limited patient
datasets [
274
]. A high-impact example includes GANs generating synthetic fMRI data to
train AI systems for diagnosing rare neurological conditions, preserving patient privacy
while improving model robustness [275].
Neuromorphic Computing for Real-Time Applications.
Neuromorphic systems, inspired by the brain’s parallel processing capabilities, will
redefine AI’s efficiency and scalability in neuroscience. These energy-efficient models will
enable the real-time processing of massive neural datasets, paving the way for applications
like adaptive BCIs and continuous monitoring during neurosurgery [276].
J. Clin. Med. 2025,14, 550 29 of 45
AI for Synthetic Biology and Brain Organoids.
The intersection of AI and synthetic biology offers unprecedented opportunities to
unravel the mysteries of the human brain. Brain organoids, lab-grown structures derived
from stem cells, replicate many features of real brain tissue, serving as invaluable models
for studying neural development, disease progression, and therapeutic responses. How-
ever, analyzing organoid data—spanning intricate gene expression patterns, structural
imaging, and dynamic electrophysiological recordings—requires advanced computational
tools [
277
]. AI-powered algorithms can help researchers efficiently process and interpret
these complex datasets, uncovering hidden patterns that reveal key insights into neural
development or the progression of neurodegenerative diseases. Generative models like
GANs could simulate organoid growth and behavior, enabling researchers to run “virtual
experiments” to predict how specific genetic mutations or drug treatments might affect
organoid functionality [
278
]. These tools not only accelerate hypothesis testing but also
reduce the need for time-intensive and costly laboratory procedures. With AI’s assis-
tance, researchers could build personalized organoids to model patient-specific conditions,
offering new pathways for understanding and treating neurological disorders [279].
Enhanced Human–Robot Interactions.
AI’s potential to improve the lives of patients extends far beyond diagnostics and
therapeutic tools, particularly through its role in advanced human–robot interactions.
Future humanoid robots, powered by AI-driven NLP and adaptive sensory systems, could
act as highly personalized assistants for individuals with neurological impairments [
280
].
For researchers, these advancements could provide invaluable support in clinical settings,
offering tools to monitor patient behavior, emotional states, and rehabilitation progress
in real time. For instance, robots could analyze subtle behavioral cues, such as changes
in facial expressions or body language, to detect stress or fatigue and provide real-time
feedback to caregivers and clinicians [281].
In research environments, AI-integrated robots could assist in experimental proce-
dures, adapting their interactions based on specific protocols or patient needs. For patients
with severe motor impairments, such as those caused by ALS or spinal cord injuries, inte-
grating AI with BCIs could enable the seamless and intuitive control of humanoid robots,
fostering independence [
282
]. Beyond direct care, these systems could play a transforma-
tive role in cognitive rehabilitation and social engagement, providing researchers with tools
to study long-term behavioral and neurological outcomes. By blending human–robot inter-
actions with neuroscience, AI is poised to redefine the boundaries of assistive technology,
offering solutions that are as functional as they are human-centered [283].
8.2. Personalized Neuroscience and Adaptive Therapies
The future of neuroscience lies in personalized approaches, where AI tailors diagnos-
tics and treatments to the unique profiles of individual patients.
Dynamic Neurotherapies.
AI will enable closed-loop systems that deliver real-time, adaptive therapies for brain
disorders. For example, deep brain stimulation (DBS) devices powered by AI will monitor
neural activity continuously, dynamically adjusting stimulation parameters to optimize
outcomes for Parkinson’s disease patients [
284
]. These adaptive systems will surpass
today’s static treatments, reducing side effects while maximizing efficacy [285].
Precision Mental Health Interventions.
AI-driven tools will integrate neuroimaging, genetic, and behavioral data to provide
hyper-personalized mental health care. For instance, AI systems could monitor changes in
brain connectivity patterns during therapy sessions, adjusting interventions for conditions
like depression or PTSD [
286
]. Highly cited research already demonstrates how ML can
J. Clin. Med. 2025,14, 550 30 of 45
predict antidepressant responses, but future systems will take this further by recommending
specific cognitive–behavioral techniques tailored to real-time patient feedback [287].
8.3. Multimodal and Temporal Data Integration
As neuroscience becomes increasingly data-driven, integrating information across
multiple modalities and time points will unlock new insights.
Unifying Diverse Data Streams.
AI systems of the future will seamlessly integrate diverse datasets, from fMRI and
EEG to genetic profiles and wearable sensor outputs. These multimodal models will
provide a more holistic understanding of brain disorders, identifying connections that are
invisible within single-modality studies [
64
]. For instance, combining EEG data with genetic
risk scores for epilepsy could predict individual seizure patterns with unprecedented
accuracy [288].
Tracking Brain Changes Over Time.
Temporal analysis will become a cornerstone of AI’s contribution to neuroscience.
Longitudinal AI models will analyze how brain structure and function evolve across the
lifespan, revealing the earliest signs of disorders like Alzheimer’s disease or schizophrenia.
By detecting these subtle changes years before symptoms arise, AI will enable preventive
interventions that preserve cognitive health [289].
8.4. Brain–Machine Interfaces: Next-Generation Applications
The integration of AI into brain–machine interfaces (BMIs) will expand their scope
from assistive technologies to tools that enhance cognition and repair neural damage.
Neural Restoration Through AI-Driven Feedback.
Future BMIs will incorporate bidirectional communication, where AI not only decodes
neural signals but also stimulates the brain to restore lost functions. For example, stroke
patients could use AI-enhanced BMIs that provide motor feedback while training the
brain’s plasticity, accelerating the recovery of movement and coordination [290,291].
AI-Enhanced Multisensory Prosthetics.
AI will drive the development of prosthetics that integrate vision, touch, and even
auditory feedback, creating a seamless experience for users. ML algorithms will optimize
these multisensory inputs in real time, allowing individuals to navigate environments and
interact with objects as naturally as possible [
292
]. Highly cited studies on tactile sensing
are already paving the way for these advancements, demonstrating how AI can simulate
lifelike sensations in prosthetic hands [293].
8.5. Cognitive Augmentation and Neuroadaptive Learning
AI’s potential extends beyond treating disorders to enhancing human cognition and
revolutionizing education.
AI-Driven Cognitive Enhancement.
AI will power tools that enhance cognitive functions like memory, attention, and
decision-making. These systems will leverage neuroplasticity principles, offering person-
alized training programs based on real-time neural feedback. For instance, AI-guided
interventions could help older adults maintain cognitive sharpness by targeting specific
neural circuits linked to age-related decline [294,295].
Neuroadaptive Learning Platforms.
Education will be transformed by AI systems that adapt to individual learners’ neural
and cognitive profiles. Neuroadaptive platforms could use wearable EEG devices to
monitor engagement and dynamically adjust lesson difficulty, ensuring that students stay in
an optimal learning zone [
296
]. These systems will be particularly impactful for individuals
with learning disabilities, offering tailored support that maximizes their potential [297].
J. Clin. Med. 2025,14, 550 31 of 45
8.6. Ethical AI in Neuroscience
As AI becomes increasingly integrated into neuroscience, addressing its ethical impli-
cations will be critical to ensuring equitable and responsible progress.
Cognitive Privacy and Data Ownership.
As AI is capable of decoding neural activity, issues of cognitive privacy will become
paramount. Future research must establish frameworks to protect individuals from the
unauthorized use of their neural data, safeguarding the fundamental right to freedom of
thought [298].
Equity and Inclusivity in AI Tools.
AI models often reflect the biases present in their training datasets, risking unequal
performance across populations. Ensuring that future AI systems are inclusive and equi-
table will require concerted efforts to collect diverse, representative data [
299
]. Initiatives
that prioritize fairness and inclusivity will help bridge healthcare disparities, ensuring that
AI benefits everyone [300].
8.7. Collaboration and Global Impact
The future of AI in neuroscience will depend on interdisciplinary collaborations and
global partnerships that accelerate innovation and expand access to advanced tools.
Interdisciplinary Research Teams.
The next phase of AI development will see the deeper integration of neuroscientists,
computer scientists, clinicians, and ethicists working together to design robust and mean-
ingful AI systems. Such collaborations will lead to tools that are scientifically rigorous,
clinically impactful, and ethically sound [301].
Global Data-Sharing Initiatives.
AI’s potential can only be fully realized through global collaboration. Federated
learning platforms will enable researchers from different regions to contribute data without
compromising privacy, fostering discoveries that reflect the diversity of global populations.
This approach will democratize access to cutting-edge AI tools, ensuring that their impact
reaches underserved communities [302].
9. Conclusions
The partnership between AI and neuroscience is unlocking new horizons in under-
standing the brain and addressing its myriad complexities. This union is far more than a
technological enhancement; it is a transformation that allows us to probe the mysteries of
the human mind, improve diagnoses, develop tailored treatments, and explore untapped
human potential. As AI continues to evolve, its role in neuroscience is becoming indis-
pensable, opening pathways that were once unimaginable. This conclusion reflects on the
journey covered in this review, highlighting key insights, the changing landscape, and the
importance of collaboration for the future.
9.1. Summary of Key Insights
The integration of AI into neuroscience has yielded remarkable advancements, funda-
mentally reshaping the field. This review has detailed several pivotal contributions:
•
Neuroimaging advancements—AI’s ability to analyze and integrate multimodal imag-
ing data has revolutionized our understanding of brain structure and function. It has
enabled the early detection of diseases like Alzheimer’s disease, mapped connectivity
disruptions in patients with psychiatric conditions, and provided detailed biomarkers
that traditional methods often missed.
•
Enhanced neural signal processing—AI has significantly improved the analysis of
neural signals from EEG, MEG, and other electrophysiological data. By leveraging
J. Clin. Med. 2025,14, 550 32 of 45
sophisticated models such as transformers and neuromorphic computing, researchers
can now decode real-time brain activity, predict seizures, and monitor dynamic neural
states with unmatched precision.
•
Personalized medicine—AI has propelled neuroscience into the era of personalized
care. Tailored diagnostics and therapies, particularly in mental health and neurodegen-
erative diseases, have shown extraordinary promise in improving treatment outcomes
and minimizing side effects.
•
Breakthroughs in brain–computer interfaces—AI-powered BCIs have moved from
experimental setups to real-world applications. These systems are restoring lost
motor and sensory functions while offering exciting possibilities for enhancing human
capabilities in ways previously thought impossible.
•
Challenges addressed, opportunities unveiled—Despite its success, AI faces challenges
such as data quality issues, interpretability, and ethical concerns. Innovations like
XAI and federated learning are paving the way for more equitable, transparent, and
reliable AI systems, ensuring broader and safer adoption.
9.2. The Evolving Landscape
The evolving relationship between AI and neuroscience is transforming the field
from reactive to proactive, enabling earlier interventions and more precise treatments.
Traditional neuroscience approaches, often limited by the complexity and volume of
data, are being replaced by AI systems capable of revealing patterns and correlations
that were previously hidden. These advancements are bridging the gap between basic
research and clinical application, revolutionizing how we diagnose, treat, and understand
brain disorders.
Beyond the clinical sphere, AI is expanding the scope of neuroscience into new ter-
ritories. Neuroadaptive learning platforms, cognitive enhancement tools, and AI-driven
neuroprosthetics are redefining what is possible in education, rehabilitation, and human
augmentation. These developments signal a future where the brain is not only studied
but actively optimized, creating opportunities to enhance cognitive function and improve
quality of life.
AI’s democratization further amplifies its impact. Through global collaborations and
equitable access initiatives, AI tools are being made available to underserved populations,
ensuring that the benefits of neuroscience reach individuals worldwide. This democratiza-
tion is fostering a more inclusive field where innovations address global health challenges
rather than being confined to privileged regions.
9.3. Final Thoughts
The integration of AI and neuroscience is one of the most exciting frontiers in science,
offering transformative potential that extends far beyond traditional boundaries. This
partnership is not just about applying advanced algorithms to complex neural datasets; it
is about reimagining how we understand and optimize the brain. Yet, the journey ahead
requires more than technological breakthroughs; it demands ethical vigilance, inclusivity,
and collaboration across disciplines.
Interdisciplinary efforts will be critical in addressing challenges such as cognitive
privacy, data fairness, and equitable access to AI-driven tools. Scientists, clinicians, tech-
nologists, and ethicists must work together to ensure that AI’s capabilities are harnessed
responsibly and for the benefit of all. Educating the next generation of professionals in the
interplay between neuroscience and AI will also be vital, preparing them to lead in this
rapidly advancing field.
J. Clin. Med. 2025,14, 550 33 of 45
Looking ahead, the possibilities are limitless. From unraveling the intricacies of brain
function to transforming patient care and enhancing human cognition, AI’s potential to
revolutionize neuroscience is boundless. This is not simply the conclusion of a review, it is
the beginning of a new era, one where the mysteries of the brain are not only unraveled but
leveraged to improve lives in profound and meaningful ways. By embracing innovation,
fostering collaboration, and prioritizing equity, we can ensure that the future of AI and
neuroscience is one of discovery, progress, and hope.
Author Contributions: Conceptualization, C.T., R.O. and C.-I.T.; investigation, M.S. and R.-A.C.-B.;
methodology, C.T., R.O., A.V.D. and M.P.R.; supervision, C.T., M.S. and C.C.; validation, C.C., C.T.,
and M.S.; visualization, C.T., M.P.R., A.V.D. and R.-A.C.-B.; writing—original draft, R.-A.C.-B., M.S.
and C.C.; writing—review and editing, M.S., C.-I.T. and R.-A.C.-B. All authors have read and agreed
to the published version of the manuscript.
Funding: The publication of this paper was supported by the University of Medicine and Pharmacy
Carol Davila through the institutional program Publish not Perish.
Conflicts of Interest: The authors declare no conflicts of interest.
References
1.
Avberšek, L.K.; Repovš, G. Deep learning in neuroimaging data analysis: Applications, challenges, and solutions. Front.
Neuroimaging 2022,1, 981642. [CrossRef] [PubMed]
2.
Badrulhisham, F.; Pogatzki-Zahn, E.; Segelcke, D.; Spisak, T.; Vollert, J. Machine learning and artificial intelligence in neuroscience:
A primer for researchers. Brain Behav. Immun. 2024,115, 470–479. [CrossRef] [PubMed]
3.
Cipollari, S.; Guarrasi, V.; Pecoraro, M.; Bicchetti, M.; Messina, E.; Farina, L.; Paci, P.; Catalano, C.; Panebianco, V. Convolutional
Neural Networks for Automated Classification of Prostate Multiparametric Magnetic Resonance Imaging Based on Image Quality.
J. Magn. Reson. Imaging JMRI 2022,55, 480–490. [CrossRef] [PubMed]
4.
Qasim Abbas, S.; Chi, L.; Chen, Y.-P.P. Transformed domain convolutional neural network for Alzheimer’s disease diagnosis
using structural MRI. Pattern Recognit. 2023,133, 109031. [CrossRef]
5.
Kaiser, J.; Mostafa, H.; Neftci, E. Synaptic Plasticity Dynamics for Deep Continuous Local Learning (DECOLLE). Front. Neurosci.
2020,14, 424. [CrossRef]
6.
Surianarayanan, C.; Lawrence, J.J.; Chelliah, P.R.; Prakash, E.; Hewage, C. Convergence of Artificial Intelligence and Neuroscience
towards the Diagnosis of Neurological Disorders—A Scoping Review. Sensors 2023,23, 3062. [CrossRef]
7.
Brandman, D.M.; Cash, S.S.; Hochberg, L.R. Review: Human intracortical recording and neural decoding for brain computer
interfaces. IEEE Trans. Neural Syst. Rehabil. Eng. 2017,25, 1687. [CrossRef]
8.
Asgher, U.; Ayaz, Y.; Taiar, R. Editorial: Advances in artificial intelligence (AI) in brain computer interface (BCI) and Industry 4.0
for human machine interaction (HMI). Front. Hum. Neurosci. 2023,17, 1320536. [CrossRef]
9.
Paul, D.; Sanap, G.; Shenoy, S.; Kalyane, D.; Kalia, K.; Tekade, R.K. Artificial intelligence in drug discovery and development.
Drug Discov. Today 2020,26, 80. [CrossRef]
10. Ilan, Y. Making use of noise in biological systems. Prog. Biophys. Mol. Biol. 2023,178, 83–90. [CrossRef]
11.
Jwa, A.S.; Poldrack, R.A. Addressing privacy risk in neuroscience data: From data protection to harm prevention. J. Law Biosci.
2022,9, lsac025. [CrossRef]
12.
Rudroff, T. Artificial Intelligence’s Transformative Role in Illuminating Brain Function in Long COVID Patients Using PET/FDG.
Brain Sci. 2024,14, 73. [CrossRef] [PubMed]
13.
Kalani, M.; Anjankar, A. Revolutionizing Neurology: The Role of Artificial Intelligence in Advancing Diagnosis and Treatment.
Cureus 2024,16, e61706. [CrossRef] [PubMed]
14.
Macpherson, T.; Churchland, A.; Sejnowski, T.; DiCarlo, J.; Kamitani, Y.; Takahashi, H.; Hikida, T. Natural and Artificial
Intelligence: A brief introduction to the interplay between AI and neuroscience research. Neural Netw. 2021,144, 603–613.
[CrossRef] [PubMed]
15.
Mesmari, S.A. Transforming Data into Actionable Insights with Cognitive Computing and AI. J. Softw. Eng. Appl. 2023,16,
211–222. [CrossRef]
16.
Hassabis, D.; Kumaran, D.; Summerfield, C.; Botvinick, M. Neuroscience-Inspired Artificial Intelligence. Neuron 2017,95, 245–258.
[CrossRef]
17.
Hu, G.; Li, H.; Zhao, W.; Hao, Y.; Bai, Z.; Nickerson, L.D.; Cong, F. Discovering hidden brain network responses to naturalistic
stimuli via tensor component analysis of multi-subject fMRI data. NeuroImage 2022,255, 119193. [CrossRef]
J. Clin. Med. 2025,14, 550 34 of 45
18.
Yuan, B.; Yang, D.; Rothberg, B.E.G.; Chang, H.; Xu, T. Unsupervised and supervised learning with neural network for human
transcriptome analysis and cancer diagnosis. Sci. Rep. 2020,10, 19106. [CrossRef]
19.
Pham, T.Q.; Matsui, T.; Chikazoe, J. Evaluation of the Hierarchical Correspondence between the Human Brain and Artificial
Neural Networks: A Review. Biology 2023,12, 1330. [CrossRef]
20.
Bernal, J.; Kushibar, K.; Asfaw, D.S.; Valverde, S.; Oliver, A.; Martí, R.; Lladó, X. Deep convolutional neural networks for brain
image analysis on magnetic resonance imaging: A review. Artif. Intell. Med. 2019,95, 64–81. [CrossRef]
21.
Khosrowshahi, F. Innovation in artificial neural network learning: Learn-On-Demand methodology. Autom. Constr. 2011,20,
1204–1210. [CrossRef]
22. Kaynak, O. The golden age of Artificial Intelligence. Discov. Artif. Intell. 2021,1, 1. [CrossRef]
23.
Yamashita, R.; Nishio, M.; Do, R.K.G.; Togashi, K. Convolutional neural networks: An overview and application in radiology.
Insights Imaging 2018,9, 611–629. [CrossRef] [PubMed]
24.
Radanliev, P. Artificial intelligence: Reflecting on the past and looking towards the next paradigm shift. J. Exp. Theor. Artif. Intell.
2024, 1–18. [CrossRef]
25.
Liu, P.; Lin, J.; Zhang, C. Heterogeneous Multivariate Functional Time Series Modeling: A State Space Approach. IEEE Trans.
Knowl. Data Eng. 2024,36, 8421–8433. [CrossRef]
26.
de Haan, E.; Padigar, M.; El Kihal, S.; Kübler, R.; Wieringa, J.E. Unstructured data research in business: Toward a structured
approach. J. Bus. Res. 2024,177, 114655. [CrossRef]
27.
Anuyah, S.; Singh, M.K.; Nyavor, H. Advancing clinical trial outcomes using deep learning and predictive modelling: Bridging
precision medicine and patient-centered care. World J. Adv. Res. Rev. 2024,24, 1–25. [CrossRef]
28.
Li, J.; Chao, N.; Li, H.; Chen, G.; Bian, S.; Wang, Z.; Ma, A. Enhancing bathymetric prediction by integrating gravity and gravity
gradient data with deep learning. Front. Mar. Sci. 2024,11, 1520401. [CrossRef]
29.
Halmi, M.R.; Putri, R.A. Implementation of Artificial Neural Network in Predicting CPO Prices Using Backpropagation. Inf. J.
Ilm. Bid. Teknol. Inf. Dan Komun. 2024,9, 161–165. [CrossRef]
30.
Orrù, G.; Piarulli, A.; Conversano, C.; Gemignani, A. Human-like problem-solving abilities in large language models using
ChatGPT. Front. Artif. Intell. 2023,6, 1199350. [CrossRef]
31.
Vrahatis, A.G.; Lazaros, K.; Kotsiantis, S. Graph Attention Networks: A Comprehensive Review of Methods and Applications.
Future Internet 2024,16, 318. [CrossRef]
32.
Zhang, G.; Liu, J.; Zhou, G.; Zhao, K.; Xie, Z.; Huang, B. Question-Directed Reasoning With Relation-Aware Graph Attention
Network for Complex Question Answering Over Knowledge Graph. IEEEACM Trans. Audio Speech Lang. Process. 2024,32,
1915–1927. [CrossRef]
33.
Xu, N.; Zhang, Y.; Du, C.; Song, J.; Huang, J.; Gong, Y.; Jiang, H.; Tong, Y.; Yin, J.; Wang, J.; et al. Prediction of Oncomelania
hupensis distribution in association with climate change using machine learning models. Parasites Vectors 2023,16, 377. [CrossRef]
[PubMed]
34.
Formica, C.; Bonanno, L.; Giambò, F.M.; Maresca, G.; Latella, D.; Marra, A.; Cucinotta, F.; Bonanno, C.; Lombardo, M.;
Tomarchio, O.; et al. Paving the Way for Predicting the Progression of Cognitive Decline: The Potential Role of Machine Learning
Algorithms in the Clinical Management of Neurodegenerative Disorders. J. Pers. Med. 2023,13, 1386. [CrossRef] [PubMed]
35.
Hebling Vieira, B.; Liem, F.; Dadi, K.; Engemann, D.A.; Gramfort, A.; Bellec, P.; Craddock, R.C.; Damoiseaux, J.S.; Steele,
C.J.; Yarkoni, T.; et al. Predicting future cognitive decline from non-brain and multimodal brain imaging data in healthy and
pathological aging. Neurobiol. Aging 2022,118, 55–65. [CrossRef]
36.
Tarahi, O.; Hamou, S.; Moufassih, M.; Agounad, S.; Azami, H.I. Decoding brain signals: A convolutional neural network approach
for motor imagery classification. E-Prime Adv. Electr. Eng. Electron. Energy 2024,7, 100451. [CrossRef]
37.
Zdorovtsova, N.; Jones, J.; Akarca, D.; Benhamou, E.; The CALM Team; Astle, D.E. Exploring neural heterogeneity in inattention
and hyperactivity. Cortex 2023,164, 90–111. [CrossRef]
38.
Haldar, D.; Kazerooni, A.F.; Arif, S.; Familiar, A.; Madhogarhia, R.; Khalili, N.; Bagheri, S.; Anderson, H.; Shaikh, I.S.;
Mahtabfar, A.; et al. Unsupervised machine learning using K-means identifies radiomic subgroups of pediatric low-grade
gliomas that correlate with key molecular markers. Neoplasia 2023,36, 100869. [CrossRef]
39.
Ma, F.; Bian, H.; Jiao, W.; Zhang, N. Single-cell RNA-seq reveals the role of YAP1 in prefrontal cortex microglia in depression.
BMC Neurol. 2024,24, 191. [CrossRef]
40.
Do, V.H.; Canzar, S. A generalization of t-SNE and UMAP to single-cell multimodal omics. Genome Biol. 2021,22, 130. [CrossRef]
41.
Loriette, C.; Amengual, J.L.; Hamed, S.B. Beyond the brain-computer interface: Decoding brain activity as a tool to understand
neuronal mechanisms subtending cognition and behavior. Front. Neurosci. 2022,16, 811736. [CrossRef] [PubMed]
42.
Avital, N.; Nahum, E.; Levi, G.C.; Malka, D. Cognitive State Classification Using Convolutional Neural Networks on Gamma-Band
EEG Signals. Appl. Sci. 2024,14, 8380. [CrossRef]
43.
Li, Z.; Chen, W.-H.; Yang, J.; Yan, Y. AID-RL: Active information-directed reinforcement learning for autonomous source seeking
and estimation. Neurocomputing 2023,544, 126281. [CrossRef]
J. Clin. Med. 2025,14, 550 35 of 45
44.
Sheng, Z.; Huang, Z.; Chen, S. Traffic expertise meets residual RL: Knowledge-informed model-based residual reinforcement
learning for CAV trajectory control. Commun. Transp. Res. 2024,4, 100142. [CrossRef]
45.
Mitjans, A.G.; Linares, D.P.; Naranjo, C.L.; Gonzalez, A.A.; Li, M.; Wang, Y.; Reyes, R.G.; Bringas-Vega, M.L.; Minati, L.;
Evans, A.C.; et al. Accurate and Efficient Simulation of Very High-Dimensional Neural Mass Models with Distributed-Delay
Connectome Tensors. NeuroImage 2023,274, 120137. [CrossRef]
46.
Górriz, J.M.; Álvarez-Illán, I.; Álvarez-Marquina, A.; Arco, J.E.; Atzmueller, M.; Ballarini, F.; Barakova, E.; Bologna, G.; Bonomini,
P.; Castellanos-Dominguez, G.; et al. Computational approaches to Explainable Artificial Intelligence: Advances in theory,
applications and trends. Inf. Fusion 2023,100, 101945. [CrossRef]
47.
Akrami, H.; Cui, W.; Kim, P.E.; Heck, C.N.; Irimia, A.; Jerbi, K.; Nair, D.; Leahy, R.M.; Joshi, A.A. Prediction of Post Traumatic
Epilepsy Using MR-Based Imaging Markers. Hum. Brain Mapp. 2024,45, e70075. [CrossRef]
48.
Martinez-Saito, M.; Gorina, E. Learning under social versus nonsocial uncertainty: A meta-analytic approach. Hum. Brain Mapp.
2022,43, 4185–4206. [CrossRef]
49.
Dalvi-Esfahani, M.; Mosharaf-Dehkordi, M.; Leong, L.W.; Ramayah, T.; Jamal Kanaan-Jebna, A.M. Exploring the drivers of
XAI-enhanced clinical decision support systems adoption: Insights from a stimulus-organism-response perspective. Technol.
Forecast. Soc. Chang. 2023,195, 122768. [CrossRef]
50.
Nasarian, E.; Alizadehsani, R.; Acharya, U.R.; Tsui, K.-L. Designing interpretable ML system to enhance trust in healthcare: A
systematic review to proposed responsible clinician-AI-collaboration framework. Inf. Fusion 2024,108, 102412. [CrossRef]
51.
Seurin, P.; Shirvan, K. Multi-objective reinforcement learning-based approach for pressurized water reactor optimization. Ann.
Nucl. Energy 2024,205, 110582. [CrossRef]
52.
Ebrie, A.S.; Kim, Y.J. Reinforcement Learning-Based Multi-Objective Optimization for Generation Scheduling in Power Systems.
Systems 2024,12, 106. [CrossRef]
53.
Bhandari, A.; Koppen, J.; Agzarian, M. Convolutional neural networks for brain tumour segmentation. Insights Imaging 2020,
11, 77. [CrossRef] [PubMed]
54.
Wang, X.; Zhang, X.; Chen, Y.; Yang, X. IFC-GNN: Combining interactions of functional connectivity with multimodal graph
neural networks for ASD brain disorder analysis. Alex. Eng. J. 2024,98, 44–55. [CrossRef]
55.
Bongiorni, L.; Balbinot, A. Evaluation of recurrent neural networks as epileptic seizure predictor. Array 2020,8, 100038. [CrossRef]
56.
Zhu, R.; Pan, W.; Liu, J.; Shang, J. Epileptic seizure prediction via multidimensional transformer and recurrent neural network
fusion. J. Transl. Med. 2024,22, 895. [CrossRef]
57.
Tunkiel, A.T.; Sui, D.; Wiktorski, T. Impact of data pre-processing techniques on recurrent neural network performance in context
of real-time drilling logs in an automated prediction framework. J. Pet. Sci. Eng. 2022,208, 109760. [CrossRef]
58.
Rakhmatulin, I.; Dao, M.-S.; Nassibi, A.; Mandic, D. Exploring Convolutional Neural Network Architectures for EEG Feature
Extraction. Sensors 2024,24, 877. [CrossRef]
59.
Brancaccio, A.; Tabarelli, D.; Zazio, A.; Bertazzoli, G.; Metsomaa, J.; Ziemann, U.; Bortoletto, M.; Belardinelli, P. Towards the
definition of a standard in TMS-EEG data preprocessing. NeuroImage 2024,301, 120874. [CrossRef]
60.
Alrawis, M.; Al-Ahmadi, S.; Mohammad, F. Bridging Modalities: A Multimodal Machine Learning Approach for Parkinson’s
Disease Diagnosis Using EEG and MRI Data. Appl. Sci. 2024,14, 3883. [CrossRef]
61.
Fan, H.; Zhang, X.; Xu, Y.; Fang, J.; Zhang, S.; Zhao, X.; Yu, J. Transformer-based multimodal feature enhancement networks
for multimodal depression detection integrating video, audio and remote photoplethysmograph signals. Inf. Fusion 2024,
104, 102161. [CrossRef]
62.
Ramzan, F.; Sartori, C.; Consoli, S.; Reforgiato Recupero, D. Generative Adversarial Networks for Synthetic Data Generation in
Finance: Evaluating Statistical Similarities and Quality Assessment. AI 2024,5, 667–685. [CrossRef]
63.
Wang, R.; Bashyam, V.; Yang, Z.; Yu, F.; Tassopoulou, V.; Chintapalli, S.S.; Skampardoni, I.; Sreepada, L.P.; Sahoo, D.;
Nikita, K.; et al. Applications of generative adversarial networks in neuroimaging and clinical neuroscience. NeuroImage
2023,269, 119898. [CrossRef] [PubMed]
64.
Alsubaie, M.G.; Luo, S.; Shaukat, K. Alzheimer’s Disease Detection Using Deep Learning on Neuroimaging: A Systematic Review.
Mach. Learn. Knowl. Extr. 2024,6, 464–505. [CrossRef]
65.
Daza, I.G.; Izquierdo, R.; Martínez, L.M.; Benderius, O.; Llorca, D.F. Sim-to-real transfer and reality gap modeling in model
predictive control for autonomous driving. Appl. Intell. 2023,53, 12719–12735. [CrossRef]
66.
Tiboni, G.; Arndt, K.; Kyrki, V. DROPO: Sim-to-real transfer with offline domain randomization. Robot. Auton. Syst. 2023,
166, 104432. [CrossRef]
67.
Seoni, S.; Jahmunah, V.; Salvi, M.; Barua, P.D.; Molinari, F.; Acharya, U.R. Application of uncertainty quantification to artificial
intelligence in healthcare: A review of last decade (2013–2023). Comput. Biol. Med. 2023,165, 107441. [CrossRef]
68.
Al Fahoum, A.; Zyout, A. Early detection of neurological abnormalities using a combined phase space reconstruction and deep
learning approach. Intell.-Based Med. 2023,8, 100123. [CrossRef]
J. Clin. Med. 2025,14, 550 36 of 45
69.
Pezoulas, V.C.; Zaridis, D.I.; Mylona, E.; Androutsos, C.; Apostolidis, K.; Tachos, N.S.; Fotiadis, D.I. Synthetic data generation
methods in healthcare: A review on open-source tools and methods. Comput. Struct. Biotechnol. J. 2024,23, 2892–2910. [CrossRef]
70.
Chen, C.; Wang, H.; Chen, Y.; Yin, Z.; Yang, X.; Ning, H.; Zhang, Q.; Li, W.; Xiao, R.; Zhao, J. Understanding the brain with
attention: A survey of transformers in brain sciences. Brain-X 2023,1, e29. [CrossRef]
71.
Basso, D.; Cottini, M. Cognitive Neuroscience and Education: Not a Gap to Be Bridged but a Common Field to Be Cultivated.
Sustainability 2023,15, 1628. [CrossRef]
72. Südhof, T.C. Molecular Neuroscience in the 21st Century: A Personal Perspective. Neuron 2017,96, 536. [CrossRef] [PubMed]
73.
Stogsdill, J.A.; Eroglu, C. The interplay between neurons and glia in synapse development and plasticity. Curr. Opin. Neurobiol.
2016,42, 1. [CrossRef] [PubMed]
74.
Bernardinelli, Y.; Nikonenko, I.; Muller, D. Structural plasticity: Mechanisms and contribution to developmental psychiatric
disorders. Front. Neuroanat. 2014,8, 123. [CrossRef] [PubMed]
75.
Wiera, G.; Jabło ´nska, J.; Lech, A.M.; Mozrzymas, J.W. Input specificity of NMDA-dependent GABAergic plasticity in the
hippocampus. Sci. Rep. 2024,14, 20463. [CrossRef]
76.
Fukaya, R.; Hirai, H.; Sakamoto, H.; Hashimotodani, Y.; Hirose, K.; Sakaba, T. Increased vesicle fusion competence underlies
long-term potentiation at hippocampal mossy fiber synapses. Sci. Adv. 2023,9, eadd3616. [CrossRef]
77.
Desai, N.S.; Zhong, C.; Kim, R.; Talmage, D.A.; Role, L.W. A simple MATLAB toolbox for analyzing calcium imaging data
in vitro
and in vivo. J. Neurosci. Methods 2024,409, 110202. [CrossRef]
78.
Lee, C.; Lee, B.H.; Jung, H.; Lee, C.; Sung, Y.; Kim, H.; Kim, J.; Shim, J.Y.; Kim, J.; Choi, D.I.; et al. Hippocampal engram networks
for fear memory recruit new synapses and modify pre-existing synapses in vivo. Curr. Biol. 2023,33, 507–516.e3. [CrossRef]
79.
Furukawa, R.; Kume, K.; Tateno, T. Analyzing the transient response dynamics of long-term depression in the mouse auditory
cortex in vitro through multielectrode-array-based spatiotemporal recordings. Front. Neurosci. 2024,18, 1448365. [CrossRef]
80.
Azimzadeh, M.; Mohd Azmi, M.A.N.; Reisi, P.; Cheah, P.-S.; Ling, K.-H. Step-by-step approach: Stereotaxic surgery for
in vivo
extracellular field potential recording at the rat Schaffer collateral-CA1 synapse using the eLab system. MethodsX 2024,12, 102544.
[CrossRef]
81.
Emery, B.A.; Hu, X.; Khanzada, S.; Kempermann, G.; Amin, H. High-resolution CMOS-based biosensor for assessing hippocampal
circuit dynamics in experience-dependent plasticity. Biosens. Bioelectron. 2023,237, 115471. [CrossRef] [PubMed]
82.
Mäki-Marttunen, T.; Blackwell, K.T.; Akkouh, I.; Shadrin, A.; Valstad, M.; Elvsåshagen, T.; Linne, M.-L.; Djurovic, S.; Einevoll, G.T.;
Andreassen, O.A. Genetic mechanisms for impaired synaptic plasticity in schizophrenia revealed by computational modeling.
Proc. Natl. Acad. Sci. USA 2024,121, e2312511121. [CrossRef] [PubMed]
83.
Petzi, M.; Singh, S.; Trappenberg, T.; Nunes, A. Mechanisms of Sustained Increases in
γ
Power Post-Ketamine in a Computational
Model of the Hippocampal CA3: Implications for Ketamine’s Antidepressant Mechanism of Action. Brain Sci. 2023,13, 1562.
[CrossRef] [PubMed]
84.
Rygvold, T.W.; Hatlestad-Hall, C.; Elvsåshagen, T.; Moberget, T.; Andersson, S. Long term potentiation-like neural plasticity and
performance-based memory function. Neurobiol. Learn. Mem. 2022,196, 107696. [CrossRef]
85.
Schimmelpfennig, J.; Topczewski, J.; Zajkowski, W.; Jankowiak-Siuda, K. The role of the salience network in cognitive and
affective deficits. Front. Hum. Neurosci. 2023,17, 1133367. [CrossRef]
86.
Logan, R.W.; McClung, C.A. Rhythms of life: Circadian disruption and brain disorders across the lifespan. Nat. Rev. Neurosci.
2019,20, 49–65. [CrossRef]
87.
Laumann, T.O.; Snyder, A.Z.; Gratton, C. Challenges in the measurement and interpretation of dynamic functional connectivity.
Imaging Neurosci. 2024,2, 1–19. [CrossRef]
88.
Yun, S.D.; Oh, S.S.; Chang, M.C. Editorial: Novel fMRI techniques and analysis methods for enhanced detection of functional
disorders. Front. Neurosci. 2024,18, 1466071. [CrossRef]
89.
Lawn, T.; Howard, M.A.; Turkheimer, F.; Misic, B.; Deco, G.; Martins, D.; Dipasquale, O. From neurotransmitters to networks:
Transcending organisational hierarchies with molecular-informed functional imaging. Neurosci. Biobehav. Rev. 2023,150, 105193.
[CrossRef]
90.
Malekian, V.; Graedel, N.N.; Hickling, A.; Aghaeifar, A.; Dymerska, B.; Corbin, N.; Josephs, O.; Maguire, E.A.; Callaghan, M.F.
Mitigating susceptibility-induced distortions in high-resolution 3DEPI fMRI at 7T. NeuroImage 2023,279, 120294. [CrossRef]
91.
Khan, A.F.; Iturria-Medina, Y. Beyond the usual suspects: Multi-factorial computational models in the search for neurodegenera-
tive disease mechanisms. Transl. Psychiatry 2024,14, 386. [CrossRef] [PubMed]
92.
Mijalkov, M.; Volpe, G.; Pereira, J.B. Directed Brain Connectivity Identifies Widespread Functional Network Abnormalities in
Parkinson’s Disease. Cereb. Cortex 2022,32, 593–607. [CrossRef] [PubMed]
93.
Anderson, L.; De Ridder, D.; Glue, P.; Mani, R.; van Sleeuwen, C.; Smith, M.; Adhia, D.B. A safety and feasibility randomized
placebo controlled trial exploring electroencephalographic effective connectivity neurofeedback treatment for fibromyalgia.
Sci. Rep. 2025,15, 209. [CrossRef]
J. Clin. Med. 2025,14, 550 37 of 45
94.
Cho, S.; van Es, M.; Woolrich, M.; Gohil, C. Comparison between EEG and MEG of static and dynamic resting-state networks.
Hum. Brain Mapp. 2024,45, e70018. [CrossRef]
95.
Blanco, R.; Koba, C.; Crimi, A. Investigating the interaction between EEG and fNIRS: A multimodal network analysis of brain
connectivity. J. Comput. Sci. 2024,82, 102416. [CrossRef]
96.
Zigmantovich, A.S.; Sharova, E.V.; Kopachka, M.M.; Smirnov, A.S.; Alexandrova, E.V.; Masherov, E.L.; Troshina, E.M.; Pronin, I.N.;
Oknina, L.B. Connectivity of EEG and fMRI Network in the Resting State in Healthy People and Patients with Post-Traumatic
Disorder of Consciousness. Hum. Physiol. 2024,50, 1–14. [CrossRef]
97.
Luo, Q.; Pan, B.; Gu, H.; Simmonite, M.; Francis, S.; Liddle, P.F.; Palaniyappan, L. Effective connectivity of the right anterior insula
in schizophrenia: The salience network and task-negative to task-positive transition. NeuroImage Clin. 2020,28, 102377. [CrossRef]
98.
Weiner, O.M.; O’Byrne, J.; Cross, N.E.; Giraud, J.; Tarelli, L.; Yue, V.; Homer, L.; Walker, K.; Carbone, R.; Dang-Vu, T.T. Slow
oscillation-spindle cross-frequency coupling predicts overnight declarative memory consolidation in older adults. Eur. J. Neurosci.
2024,59, 662–685. [CrossRef]
99.
Wang, Y.; Chen, H.; Wang, C.; Liu, J.; Miao, P.; Wei, Y.; Wu, L.; Wang, X.; Wang, P.; Zhang, Y.; et al. Static and dynamic interactions
within the triple-network model in stroke patients with multidomain cognitive impairments. NeuroImage Clin. 2024,43, 103655.
[CrossRef]
100.
Figueroa-Vargas, A.; Góngora, B.; Alonso, M.F.; Ortega, A.; Soto-Fernández, P.; Z-Rivera, L.; Ramírez, S.; González, F.; Muñoz
Venturelli, P.; Billeke, P. The effect of a cognitive training therapy based on stimulation of brain oscillations in patients with
mild cognitive impairment in a Chilean sample: Study protocol for a phase IIb, 2
×
3 mixed factorial, double-blind randomised
controlled trial. Trials 2024,25, 144. [CrossRef]
101.
Widge, A.S.; Miller, E.K. Targeting Cognition and Networks Through Neural Oscillations: Next-Generation Clinical Brain
Stimulation. JAMA Psychiatry 2019,76, 671. [CrossRef] [PubMed]
102.
Zrenner, C. Brain-oscillation phase as a therapeutic target for EEG-synchronized TMS. Brain Stimul. Basic Transl. Clin. Res.
Neuromodul. 2023,16, 158. [CrossRef]
103.
Cuevas-Diaz Duran, R.; Wei, H.; Wu, J.Q. Single-cell RNA-sequencing of the brain. Clin. Transl. Med. 2017,6, 20. [CrossRef]
[PubMed]
104.
Piwecka, M.; Rajewsky, N.; Rybak-Wolf, A. Single-cell and spatial transcriptomics: Deciphering brain complexity in health and
disease. Nat. Rev. Neurol. 2023,19, 346–362. [CrossRef]
105.
Turner, B.M.; Rodriguez, C.A.; Norcia, T.M.; McClure, S.M.; Steyvers, M. Why more is better: Simultaneous modeling of EEG,
fMRI, and behavioral data. NeuroImage 2016,128, 96–115. [CrossRef]
106.
Schurz, M.; Uddin, L.Q.; Kanske, P.; Lamm, C.; Sallet, J.; Bernhardt, B.C.; Mars, R.B.; Bzdok, D. Variability in Brain Structure and
Function Reflects Lack of Peer Support. Cereb. Cortex 2021,31, 4612–4627. [CrossRef]
107.
Leve, L.D.; Kanamori, M.; Humphreys, K.L.; Jaffee, S.R.; Nusslock, R.; Oro, V.; Hyde, L.W. The Promise and Challenges of
Integrating Biological and Prevention Sciences: A Community-Engaged Model for the Next Generation of Translational Research.
Prev. Sci. 2024,25, 1177–1199. [CrossRef]
108.
Livanis, E.; Voultsos, P.; Vadikolias, K.; Pantazakos, P.; Tsaroucha, A. Understanding the Ethical Issues of Brain-Computer
Interfaces (BCIs): A Blessing or the Beginning of a Dystopian Future? Cureus 2024,16, e58243. [CrossRef]
109.
Dipietro, L.; Gonzalez-Mego, P.; Ramos-Estebanez, C.; Zukowski, L.H.; Mikkilineni, R.; Rushmore, R.J.; Wagner, T. The evolution
of Big Data in neuroscience and neurology. J. Big Data 2023,10, 116. [CrossRef]
110.
Wu, D.; Faria, A.V.; Younes, L.; Mori, S.; Brown, T.; Johnson, H.; Paulsen, J.S.; Ross, C.A.; Miller, M.I.; PREDICT-HD Investigators
and Coordinators of the Huntington Study Group. Mapping the order and pattern of brain structural MRI changes using
change-point analysis in premanifest Huntington’s disease. Hum. Brain Mapp. 2017,38, 5035. [CrossRef]
111.
Aberathne, I.; Kulasiri, D.; Samarasinghe, S. Detection of Alzheimer’s disease onset using MRI and PET neuroimaging: Longitu-
dinal data analysis and machine learning. Neural Regen. Res. 2023,18, 2134. [CrossRef] [PubMed]
112.
Stark, A.J.; Claassen, D.O. Positron emission tomography in Parkinson’s disease: Insights into impulsivity. Int. Rev. Psychiatry
Abingdon Engl. 2017,29, 618–627. [CrossRef] [PubMed]
113.
Stokes, M.G.; Wolff, M.J.; Spaak, E. Decoding Rich Spatial Information with High Temporal Resolution. Trends Cogn. Sci. 2015,
19, 636. [CrossRef] [PubMed]
114.
Flores, A.; Münte, T.F.; Doñamayor, N. Event-related EEG responses to anticipation and delivery of monetary and social reward.
Biol. Psychol. 2015,109, 10–19. [CrossRef]
115.
Yao, X.; Glessner, J.T.; Li, J.; Qi, X.; Hou, X.; Zhu, C.; Li, X.; March, M.E.; Yang, L.; Mentch, F.D.; et al. Integrative analysis of
genome-wide association studies identifies novel loci associated with neuropsychiatric disorders. Transl. Psychiatry 2021,11, 69.
[CrossRef]
116.
Bareši´c, A.; Nash, A.J.; Dahoun, T.; Howes, O.; Lenhard, B. Understanding the genetics of neuropsychiatric disorders: The
potential role of genomic regulatory blocks. Mol. Psychiatry 2020,25, 6–18. [CrossRef]
117. Johnson, K.T.; Picard, R.W. Advancing Neuroscience through Wearable Devices. Neuron 2020,108, 8–12. [CrossRef]
J. Clin. Med. 2025,14, 550 38 of 45
118.
Riva, G.; Wiederhold, B.K.; Mantovani, F. Neuroscience of Virtual Reality: From Virtual Exposure to Embodied Medicine.
Cyberpsychol. Behav. Soc. Netw. 2019,22, 82. [CrossRef]
119.
Prevedel, R.; Verhoef, A.J.; Pernia-Andrade, A.J.; Weisenburger, S.; Huang, B.S.; Nöbauer, T.; Fernández, A.; Delcour, J.E.; Golshani,
P.; Baltuska, A.; et al. Fast volumetric calcium imaging across multiple cortical layers using sculpted light. Nat. Methods 2016,
13, 1021. [CrossRef]
120.
Epp, J.R.; Niibori, Y.; Hsiang, H.-L.; Mercaldo, V.; Deisseroth, K.; Josselyn, S.A.; Frankland, P.W. Optimization of CLARITY for
Clearing Whole-Brain and Other Intact Organs. eNeuro 2015,2, 1–15. [CrossRef]
121.
Han, X.; Cramer, S.R.; Zhang, N. Deriving causal relationships in resting-state functional connectivity using SSFO-based
optogenetic fMRI. J. Neural Eng. 2022,19, 066002. [CrossRef] [PubMed]
122.
Lee, G.; Nho, K.; Kang, B.; Sohn, K.-A.; Kim, D. Predicting Alzheimer’s disease progression using multi-modal deep learning
approach. Sci. Rep. 2019,9, 1952. [CrossRef] [PubMed]
123.
Yang, G.R.; Wang, X.-J. Artificial Neural Networks for Neuroscientists: A Primer. Neuron 2020,107, 1048–1070. [CrossRef]
[PubMed]
124.
Lim, S. Hebbian learning revisited and its inference underlying cognitive function. Curr. Opin. Behav. Sci. 2021,38, 96–102.
[CrossRef]
125.
Krotov, D.; Hopfield, J.J. Unsupervised learning by competing hidden units. Proc. Natl. Acad. Sci. USA 2019,116, 7723–7731.
[CrossRef]
126.
Saponati, M.; Vinck, M. Sequence anticipation and spike-timing-dependent plasticity emerge from a predictive learning rule. Nat.
Commun. 2023,14, 4985. [CrossRef]
127. Roudi, Y.; Taylor, G. Learning with hidden variables. Curr. Opin. Neurobiol. 2015,35, 110–118. [CrossRef]
128. Espinosa, J.S.; Stryker, M.P. Development and Plasticity of the Primary Visual Cortex. Neuron 2012,75, 230. [CrossRef]
129.
Celeghin, A.; Borriero, A.; Orsenigo, D.; Diano, M.; Guerrero, C.A.M.; Perotti, A.; Petri, G.; Tamietto, M. Convolutional neural
networks for vision neuroscience: Significance, developments, and outstanding issues. Front. Comput. Neurosci. 2023,17, 1153572.
[CrossRef]
130.
Gopinath, N. Artificial intelligence and neuroscience: An update on fascinating relationships. Process Biochem. 2023,125, 113–120.
[CrossRef]
131.
Calhoun, V.D.; Pearlson, G.D.; Sui, J. Data-driven approaches to neuroimaging biomarkers for neurological and psychiatric
disorders: Emerging approaches and examples. Curr. Opin. Neurol. 2021,34, 469. [CrossRef] [PubMed]
132.
Muñoz-Ramírez, V.; Kmetzsch, V.; Forbes, F.; Meoni, S.; Moro, E.; Dojat, M. Subtle anomaly detection: Application to brain MRI
analysis of de novo Parkinsonian patients. Artif. Intell. Med. 2022,125, 102251. [CrossRef] [PubMed]
133.
Golshanrad, P.; Faghih, F. DeepCover: Advancing RNN test coverage and online error prediction using state machine extraction.
J. Syst. Softw. 2024,211, 111987. [CrossRef]
134.
Ein Shoka, A.A.; Dessouky, M.M.; El-Sayed, A.; Hemdan, E.E.-D. EEG seizure detection: Concepts, techniques, challenges, and
future trends. Multimed. Tools Appl. 2023,82, 42021–42051. [CrossRef]
135.
Gunasekaran, K.; Ambeth Kumar, V.D. Artifact removal from ECG signals using online recursive independent component
analysis. J. Comput. Math. Data Sci. 2024,13, 100102. [CrossRef]
136.
Abdelfattah, T.; Maher, A.; Youssef, A.; Driessen, P.F. Seamless Optimization of Wavelet Parameters for Denoising LFM Radar
Signals: An AI-Based Approach. Remote Sens. 2024,16, 4211. [CrossRef]
137.
Fang, J.; Guo, X.; Liu, Y.; Chang, X.; Fujita, H.; Wu, J. An attention-based deep learning model for multi-horizon time series
forecasting by considering periodic characteristic. Comput. Ind. Eng. 2023,185, 109667. [CrossRef]
138. Nguyen, H.A.T.; Pham, D.H.; Ahn, Y. Effect of Data Augmentation Using Deep Learning on Predictive Models for Geopolymer
Compressive Strength. Appl. Sci. 2024,14, 3601. [CrossRef]
139.
Torres-Gaona, G.; Aledo-Serrano, Á.; García-Morales, I.; Toledano, R.; Valls, J.; Cosculluela, B.; Munsó, L.; Raurich, X.; Trejo, A.;
Blanquez, D.; et al. Artificial intelligence system, based on mjn-SERAS algorithm, for the early detection of seizures in patients
with refractory focal epilepsy: A cross-sectional pilot study. Epilepsy Behav. Rep. 2023,22, 100600. [CrossRef]
140.
Statsenko, Y.; Babushkin, V.; Talako, T.; Kurbatova, T.; Smetanina, D.; Simiyu, G.L.; Habuza, T.; Ismail, F.; Almansoori, T.M.;
Gorkom, K.N.-V.; et al. Automatic Detection and Classification of Epileptic Seizures from EEG Data: Finding Optimal Acquisition
Settings and Testing Interpretable Machine Learning Approach. Biomedicines 2023,11, 2370. [CrossRef]
141.
Yu, S.; El Atrache, R.; Tang, J.; Jackson, M.; Makarucha, A.; Cantley, S.; Sheehan, T.; Vieluf, S.; Zhang, B.; Rogers, J.L.; et al.
Artificial intelligence-enhanced epileptic seizure detection by wearables. Epilepsia 2023,64, 3213–3226. [CrossRef] [PubMed]
142.
Zhang, X.; Ma, Z.; Zheng, H.; Li, T.; Chen, K.; Wang, X.; Liu, C.; Xu, L.; Wu, X.; Lin, D.; et al. The combination of brain-computer
interfaces and artificial intelligence: Applications and challenges. Ann. Transl. Med. 2020,8, 712. [CrossRef] [PubMed]
143.
Zhao, X.; Wu, J.; Peng, H.; Beheshti, A.; Monaghan, J.J.M.; McAlpine, D.; Hernandez-Perez, H.; Dras, M.; Dai, Q.; Li, Y.; et al. Deep
reinforcement learning guided graph neural networks for brain network analysis. Neural Netw. 2022,154, 56–67. [CrossRef]
J. Clin. Med. 2025,14, 550 39 of 45
144.
Adhau, S.; Gros, S.; Skogestad, S. Reinforcement learning based MPC with neural dynamical models. Eur. J. Control. 2024,
80, 101048. [CrossRef]
145.
Alayba, A.M.; Senan, E.M.; Alshudukhi, J.S. Enhancing early detection of Alzheimer’s disease through hybrid models based on
feature fusion of multi-CNN and handcrafted features. Sci. Rep. 2024,14, 31203. [CrossRef]
146.
Al-Ezzi, A.; Arechavala, R.J.; Butler, R.; Nolty, A.; Kang, J.J.; Shimojo, S.; Wu, D.-A.; Fonteh, A.N.; Kleinman, M.T.;
Kloner, R.A.; et al. Disrupted brain functional connectivity as early signature in cognitively healthy individuals with pathological
CSF amyloid/tau. Commun. Biol. 2024,7, 1037. [CrossRef]
147.
Wang, S.-M.; Kang, D.W.; Um, Y.H.; Kim, S.; Lee, C.U.; Lim, H.K. Depression Is Associated with the Aberration of Resting State
Default Mode Network Functional Connectivity in Patients with Amyloid-Positive Mild Cognitive Impairment. Brain Sci. 2023,
13, 1111. [CrossRef]
148.
Toumaj, S.; Heidari, A.; Shahhosseini, R.; Jafari Navimipour, N. Applications of deep learning in Alzheimer’s disease: A
systematic literature review of current trends, methodologies, challenges, innovations, and future directions. Artif. Intell. Rev.
2024,58, 44. [CrossRef]
149. Xu, M.; Ouyang, Y.; Yuan, Z. Deep Learning Aided Neuroimaging and Brain Regulation. Sensors 2023,23, 4993. [CrossRef]
150.
Newby, D.; Orgeta, V.; Marshall, C.R.; Lourida, I.; Albertyn, C.P.; Tamburin, S.; Raymont, V.; Veldsman, M.; Koychev, I.;
Bauermeister, S.; et al. Artificial Intelligence for Dementia Prevention. Alzheimers Dement. J. Alzheimers Assoc. 2023,19, 5952–5969.
[CrossRef]
151.
Castellano, G.; Esposito, A.; Lella, E.; Montanaro, G.; Vessio, G. Automated detection of Alzheimer’s disease: A multi-modal
approach with 3D MRI and amyloid PET. Sci. Rep. 2024,14, 5210. [CrossRef] [PubMed]
152.
Canny, E.; Vansteensel, M.J.; van der Salm, S.M.A.; Müller-Putz, G.R.; Berezutskaya, J. Boosting brain–computer interfaces with
functional electrical stimulation: Potential applications in people with locked-in syndrome. J. NeuroEng. Rehabil. 2023,20, 157.
[CrossRef] [PubMed]
153.
de Filippis, R.; Foysal, A.A. Chatbots in Psychology: Revolutionizing Clinical Support and Mental Health Care. Voice Publ. 2024,
10, 298–321. [CrossRef]
154.
van der Schyff, E.L.; Ridout, B.; Amon, K.L.; Forsyth, R.; Campbell, A.J. Providing Self-Led Mental Health Support Through an
Artificial Intelligence–Powered Chat Bot (Leora) to Meet the Demand of Mental Health Care. J. Med. Internet Res. 2023,25, e46448.
[CrossRef] [PubMed]
155.
Soleimani Rad, H.; Goodarzi, H.; Bahrami, L.; Abolghasemi, A. Internet-Based Versus Face-to-Face Cognitive-Behavioral Therapy
for Social Anxiety Disorder: A Randomized Control Trial. Behav. Ther. 2024,55, 528–542. [CrossRef]
156.
Vrahatis, A.G.; Skolariki, K.; Krokidis, M.G.; Lazaros, K.; Exarchos, T.P.; Vlamos, P. Revolutionizing the Early Detection of
Alzheimer’s Disease through Non-Invasive Biomarkers: The Role of Artificial Intelligence and Deep Learning. Sensors 2023,
23, 4184. [CrossRef]
157.
Maleki Varnosfaderani, S.; Forouzanfar, M. The Role of AI in Hospitals and Clinics: Transforming Healthcare in the 21st Century.
Bioengineering 2024,11, 337. [CrossRef]
158.
Kulkarni, A.; Anderson, A.G.; Merullo, D.P.; Konopka, G. Beyond bulk: A review of single cell transcriptomics methodologies
and applications. Curr. Opin. Biotechnol. 2019,58, 129. [CrossRef]
159.
Mishra, R.; Li, B. The Application of Artificial Intelligence in the Genetic Study of Alzheimer’s Disease. Aging Dis. 2020,11, 1567.
[CrossRef]
160.
Imoto, Y.; Nakamura, T.; Escolar, E.G.; Yoshiwaki, M.; Kojima, Y.; Yabuta, Y.; Katou, Y.; Yamamoto, T.; Hiraoka, Y.; Saitou, M.
Resolution of the curse of dimensionality in single-cell RNA sequencing data analysis. Life Sci. Alliance 2022,5, e202201591.
[CrossRef]
161.
Al-Sakkari, E.G.; Ragab, A.; Amer, M.; Ajao, O.; Benali, M.; Boffito, D.C.; Dagdougui, H.; Amazouz, M. Ensemble machine
learning to accelerate industrial decarbonization: Prediction of Hansen solubility parameters for streamlined chemical solvent
selection. Digit. Chem. Eng. 2025,14, 100207. [CrossRef]
162.
Yu, Y.; Mai, Y.; Zheng, Y.; Shi, L. Assessing and mitigating batch effects in large-scale omics studies. Genome Biol. 2024,25, 254.
[CrossRef] [PubMed]
163. Arevalo, J.; Su, E.; Ewald, J.D.; van Dijk, R.; Carpenter, A.E.; Singh, S. Evaluating batch correction methods for image-based cell
profiling. Nat. Commun. 2024,15, 6516. [CrossRef] [PubMed]
164.
Marey, A.; Arjmand, P.; Alerab, A.D.S.; Eslami, M.J.; Saad, A.M.; Sanchez, N.; Umair, M. Explainability, transparency and black
box challenges of AI in radiology: Impact on patient care in cardiovascular radiology. Egypt. J. Radiol. Nucl. Med. 2024,55, 183.
[CrossRef]
165.
Bouchard, C.; Bernatchez, R.; Lavoie-Cardinal, F. Addressing annotation and data scarcity when designing machine learning
strategies for neurophotonics. Neurophotonics 2023,10, 044405. [CrossRef]
166.
Rathi, N.; Chakraborty, I.; Kosta, A.; Sengupta, A.; Ankit, A.; Panda, P.; Roy, K. Exploring Neuromorphic Computing Based on
Spiking Neural Networks: Algorithms to Hardware. ACM Comput. Surv. 2023,55, 1–49. [CrossRef]
J. Clin. Med. 2025,14, 550 40 of 45
167.
Prakash, C.; Raj Gupta, L.; Mehta, A.; Vasudev, H.; Tominov, R.; Korman, E.; Fedotov, A.; Smirnov, V.; Kumar Kesari, K.
Computing of neuromorphic materials: An emerging approach for bioengineering solutions. Mater. Adv. 2023,4, 5882–5919.
[CrossRef]
168.
Zhao, L.; Zhang, L.; Wu, Z.; Chen, Y.; Dai, H.; Yu, X.; Liu, Z.; Zhang, T.; Hu, X.; Jiang, X.; et al. When brain-inspired AI meets AGI.
Meta-Radiol. 2023,1, 100005. [CrossRef]
169.
Ahmedov, H.B.; Yi, D.; Sui, J. Application of a brain-inspired deep imitation learning algorithm in autonomous driving. Softw.
Impacts 2021,10, 100165. [CrossRef]
170.
Seyyedabbasi, A. A reinforcement learning-based metaheuristic algorithm for solving global optimization problems. Adv. Eng.
Softw. 2023,178, 103411. [CrossRef]
171.
Gao, Q.; Schweidtmann, A.M. Deep reinforcement learning for process design: Review and perspective. Curr. Opin. Chem. Eng.
2024,44, 101012. [CrossRef]
172.
Dohare, S.; Hernandez-Garcia, J.F.; Lan, Q.; Rahman, P.; Mahmood, A.R.; Sutton, R.S. Loss of plasticity in deep continual learning.
Nature 2024,632, 768–774. [CrossRef]
173.
Rudroff, T.; Rainio, O.; Klén, R. Neuroplasticity Meets Artificial Intelligence: A Hippocampus-Inspired Approach to the Stability–
Plasticity Dilemma. Brain Sci. 2024,14, 1111. [CrossRef]
174.
S Band, S.; Yarahmadi, A.; Hsu, C.-C.; Biyari, M.; Sookhak, M.; Ameri, R.; Dehzangi, I.; Chronopoulos, A.T.; Liang, H.-W.
Application of explainable artificial intelligence in medical health: A systematic review of interpretability methods. Inform. Med.
Unlocked 2023,40, 101286. [CrossRef]
175.
Siachos, I.; Karacapilidis, N. Explainable Artificial Intelligence Methods to Enhance Transparency and Trust in Digital Deliberation
Settings. Future Internet 2024,16, 241. [CrossRef]
176.
Brahma, N.; Vimal, S. Artificial intelligence in neuroimaging: Opportunities and ethical challenges. Brain Spine 2024,4, 102919.
[CrossRef]
177.
Qian, Y.; Alhaskawi, A.; Dong, Y.; Ni, J.; Abdalbary, S.; Lu, H. Transforming medicine: Artificial intelligence integration in the
peripheral nervous system. Front. Neurol. 2024,15, 1332048. [CrossRef]
178.
Warren, S.L.; Moustafa, A.A. Functional magnetic resonance imaging, deep learning, and Alzheimer’s disease: A systematic
review. J. Neuroimaging 2023,33, 5–18. [CrossRef]
179.
Youssef, N.; Xiao, S.; Liu, M.; Lian, H.; Li, R.; Chen, X.; Zhang, W.; Zheng, X.; Li, Y.; Li, Y. Functional Brain Networks in Mild
Cognitive Impairment Based on Resting Electroencephalography Signals. Front. Comput. Neurosci. 2021,15, 698386. [CrossRef]
180.
Cui, Y.; Zhao, S.; Wang, H.; Xie, L.; Chen, Y.; Han, J.; Guo, L.; Zhou, F.; Liu, T. Identifying Brain Networks at Multiple Time Scales
via Deep Recurrent Neural Network. IEEE J. Biomed. Health Inform. 2018,23, 2515. [CrossRef]
181.
Liang, R.; Zhang, X.; Li, Q.; Wei, L.; Liu, H.; Kumar, A.; Leadingham, K.M.K.; Punnoose, J.; Garcia, L.P.; Manbachi, A.
Unidirectional brain-computer interface: Artificial neural network encoding natural images to fMRI response in the visual cortex.
arXiv 2023, arXiv:2309.15018v1.
182.
Cui, C.; Yang, H.; Wang, Y.; Zhao, S.; Asad, Z.; Coburn, L.A.; Wilson, K.T.; Landman, B.A.; Huo, Y. Deep multimodal fusion of
image and non-image data in disease diagnosis and prognosis: A review. Prog. Biomed. Eng. 2023,5, 022001. [CrossRef] [PubMed]
183.
Luo, N.; Shi, W.; Yang, Z.; Song, M.; Jiang, T. Multimodal Fusion of Brain Imaging Data: Methods and Applications. Mach. Intell.
Res. 2024,21, 136–152. [CrossRef]
184.
Khosla, M.; Ngo, G.H.; Jamison, K.; Kuceyeski, A.; Sabuncu, M.R. Cortical response to naturalistic stimuli is largely predictable
with deep neural networks. Sci. Adv. 2021,7, eabe7547. [CrossRef]
185.
Saidi, S.; Idbraim, S.; Karmoude, Y.; Masse, A.; Arbelo, M. Deep-Learning for Change Detection Using Multi-Modal Fusion of
Remote Sensing Images: A Review. Remote Sens. 2024,16, 3852. [CrossRef]
186.
Gu, Z.; Jamison, K.; Sabuncu, M.R.; Kuceyeski, A. Modulating human brain responses via optimal natural image selection and
synthetic image generation. arXiv 2023, arXiv:2304.09225v1.
187.
Nakach, F.-Z.; Idri, A.; Goceri, E. A comprehensive investigation of multimodal deep learning fusion strategies for breast cancer
classification. Artif. Intell. Rev. 2024,57, 327. [CrossRef]
188.
Chen, X.; Xie, H.; Li, Z.; Cheng, G.; Leng, M.; Wang, F.L. Information fusion and artificial intelligence for smart healthcare: A
bibliometric study. Inf. Process. Manag. 2023,60, 103113. [CrossRef]
189.
Ranjbarzadeh, R.; Bagherian Kasgari, A.; Jafarzadeh Ghoushchi, S.; Anari, S.; Naseri, M.; Bendechache, M. Brain tumor
segmentation based on deep learning and an attention mechanism using MRI multi-modalities brain images. Sci. Rep. 2021,
11, 10930. [CrossRef]
190.
Siddhad, G.; Gupta, A.; Dogra, D.P.; Roy, P.P. Efficacy of transformer networks for classification of EEG data. Biomed. Signal
Process. Control 2024,87, 105488. [CrossRef]
191.
Abubaker, M.; Al Qasem, W.; Pilátová, K.; Ježdík, P.; Kvaš ˇnák, E. Theta-gamma-coupling as predictor of working memory
performance in young and elderly healthy people. Mol. Brain 2024,17, 74. [CrossRef] [PubMed]
J. Clin. Med. 2025,14, 550 41 of 45
192.
Tuncer, T.; Dogan, S.; Subasi, A. EEG-based driving fatigue detection using multilevel feature extraction and iterative hybrid
feature selection. Biomed. Signal Process. Control 2021,68, 102591. [CrossRef]
193.
Shen, M.; Yang, F.; Wen, P.; Song, B.; Li, Y. A real-time epilepsy seizure detection approach based on EEG using short-time Fourier
transform and Google-Net convolutional neural network. Heliyon 2024,10, e31827. [CrossRef] [PubMed]
194.
Kumar, S.; Sharma, S. An Improved Deep Learning Framework for Multimodal Medical Data Analysis. Big Data Cogn. Comput.
2024,8, 125. [CrossRef]
195.
Abduljaleel, I.Q.; Ali, I.H. Deep Learning and Fusion Mechanism-based Multimodal Fake News Detection Methodologies: A
Review. Eng. Technol. Appl. Sci. Res. 2024,14, 15665–15675. [CrossRef]
196.
Matar, M.; Estevez, P.G.; Marchi, P.; Messina, F.; Elmoudi, R.; Wshah, S. Transformer-based deep learning model for forced
oscillation localization. Int. J. Electr. Power Energy Syst. 2023,146, 108805. [CrossRef]
197.
Alessandrini, M.; Biagetti, G.; Crippa, P.; Falaschetti, L.; Luzzi, S.; Turchetti, C. EEG-Based Alzheimer’s Disease Recognition
Using Robust-PCA and LSTM Recurrent Neural Network. Sensors 2022,22, 3696. [CrossRef]
198.
Ahmadzadeh Nobari Azar, N.; Cavus, N.; Esmaili, P.; Sekeroglu, B.; A¸sır, S. Detecting emotions through EEG signals based on
modified convolutional fuzzy neural network. Sci. Rep. 2024,14, 10371. [CrossRef]
199.
Norman, S.L.; Maresca, D.; Christopoulos, V.N.; Griggs, W.S.; Demene, C.; Tanter, M.; Shapiro, M.G.; Andersen, R.A. Single-trial
decoding of movement intentions using functional ultrasound neuroimaging. Neuron 2021,109, 1554–1566.e4. [CrossRef]
200. Jirsa, V.; Müller, V. Cross-frequency coupling in real and virtual brain networks. Front. Comput. Neurosci. 2013,7, 78. [CrossRef]
201.
Assali, I.; Ghazi Blaiech, A.; Ben Abdallah, A.; Ben Khalifa, K.; Carrère, M.; Hédi Bedoui, M. CNN-based classification of epileptic
states for seizure prediction using combined temporal and spectral features. Biomed. Signal Process. Control 2023,82, 104519.
[CrossRef]
202.
Stieger, J.R.; Engel, S.A.; Suma, D.; He, B. Benefits of deep learning classification of continuous noninvasive brain–computer
interface control. J. Neural Eng. 2021,18, 046082. [CrossRef] [PubMed]
203.
Ma, Y.; Gong, A.; Nan, W.; Ding, P.; Wang, F.; Fu, Y. Personalized Brain–Computer Interface and Its Applications. J. Pers. Med.
2023,13, 46. [CrossRef] [PubMed]
204. Shih, J.J.; Krusienski, D.J.; Wolpaw, J.R. Brain-Computer Interfaces in Medicine. Mayo Clin. Proc. 2012,87, 268. [CrossRef]
205.
Lehmann, H.; Arkadir, D.; Ilan, Y. Methods for Improving Brain-Computer Interface: Using A Brain-Directed Adjuvant and A
Second-Generation Artificial Intelligence System to Enhance Information Streaming and Effectiveness of Stimuli. Int. J. Appl. Biol.
Pharm. Technol. 2023,14, 42–52. [CrossRef]
206.
Barnova, K.; Mikolasova, M.; Kahankova, R.V.; Jaros, R.; Kawala-Sterniuk, A.; Snasel, V.; Mirjalili, S.; Pelc, M.; Martinek, R.
Implementation of artificial intelligence and machine learning-based methods in brain–computer interaction. Comput. Biol. Med.
2023,163, 107135. [CrossRef]
207.
Wei, Q.; Zhang, Y.; Wang, Y.; Gao, X. A Canonical Correlation Analysis-Based Transfer Learning Framework for Enhancing the
Performance of SSVEP-Based BCIs. IEEE Trans. Neural Syst. Rehabil. Eng. 2023,31, 2809–2821. [CrossRef]
208.
Schiffer, A.-M.; Siletti, K.; Waszak, F.; Yeung, N. Adaptive behaviour and feedback processing integrate experience and instruction
in reinforcement learning. NeuroImage 2017,146, 626–641. [CrossRef]
209.
Palumbo, A.; Gramigna, V.; Calabrese, B.; Ielpo, N. Motor-Imagery EEG-Based BCIs in Wheelchair Movement and Control: A
Systematic Literature Review. Sensors 2021,21, 6285. [CrossRef]
210.
Naik, R.; Chaudhari, K.; Jadhav, K.; Joshi, A. MindCeive: Perceiving human imagination using CNN-GRU and GANs. Biomed.
Signal Process. Control 2025,100, 107110. [CrossRef]
211.
George, R.; Chiappalone, M.; Giugliano, M.; Levi, T.; Vassanelli, S.; Partzsch, J.; Mayr, C. Plasticity and Adaptation in Neuromor-
phic Biohybrid Systems. iScience 2020,23, 101589. [CrossRef] [PubMed]
212.
Maye, A.; Zhang, D.; Wang, Y.; Gao, S.; Engel, A.K. Multimodal Brain-Computer Interfaces. Tsinghua Sci. Technol. 2011,16,
133–139. [CrossRef]
213.
Torres, E.P.; Torres, E.A.; Hernández-Álvarez, M.; Yoo, S.G. EEG-Based BCI Emotion Recognition: A Survey. Sensors 2020,20, 5083.
[CrossRef] [PubMed]
214.
Moreno-Calderón, S.; Martínez-Cagigal, V.; Santamaría-Vázquez, E.; Pérez-Velasco, S.; Marcos-Martínez, D.; Hornero, R.
Combining brain-computer interfaces and multiplayer video games: An application based on c-VEPs. Front. Hum. Neurosci. 2023,
17, 1227727. [CrossRef]
215.
Vieth, M.; Rahimi, A.; Gorgan Mohammadi, A.; Triesch, J.; Ganjtabesh, M. Accelerating spiking neural network simulations with
PymoNNto and PymoNNtorch. Front. Neuroinform. 2024,18, 1331220. [CrossRef]
216.
Deng, L.; Rattadilok, P. A Sensor and Machine Learning-Based Sensory Management Recommendation System for Children with
Autism Spectrum Disorders. Sensors 2022,22, 5803. [CrossRef]
217.
Binder, J.; Ursu, O.; Bologa, C.; Jiang, S.; Maphis, N.; Dadras, S.; Chisholm, D.; Weick, J.; Myers, O.; Kumar, P.; et al. Machine
learning prediction and tau-based screening identifies potential Alzheimer’s disease genes relevant to immunity. Commun. Biol.
2022,5, 125. [CrossRef]
J. Clin. Med. 2025,14, 550 42 of 45
218.
Iturria-Medina, Y.; Carbonell, F.; Assadi, A.; Adewale, Q.; Khan, A.F.; Baumeister, T.R.; Sanchez-Rodriguez, L. Integrating
molecular, histopathological, neuroimaging and clinical neuroscience data with NeuroPM-box. Commun. Biol. 2021,4, 614.
[CrossRef]
219.
Vatansever, S.; Schlessinger, A.; Wacker, D.; Kaniskan, H.Ü.; Jin, J.; Zhou, M.-M.; Zhang, B. Artificial intelligence and machine
learning-aided drug discovery in central nervous system diseases: State-of-the-arts and future directions. Med. Res. Rev. 2020,
41, 1427. [CrossRef]
220.
Du, Y.; Clair, G.C.; Alam, D.A.; Danopoulos, S.; Schnell, D.; Kitzmiller, J.A.; Misra, R.S.; Bhattacharya, S.; Warburton, D.;
Mariani, T.J.; et al. Integration of transcriptomic and proteomic data identifies biological functions in cell populations from
human infant lung. Am. J. Physiol. Lung Cell. Mol. Physiol. 2019,317, L347. [CrossRef]
221.
Zheng, Y.; Zhang, L.; Zhao, J.; Li, L.; Wang, M.; Gao, P.; Wang, Q.; Zhang, X.; Wang, W. Advances in aptamers against A
β
and
applications in Aβdetection and regulation for Alzheimer’s disease. Theranostics 2022,12, 2095. [CrossRef] [PubMed]
222.
Kokudeva, M.; Vichev, M.; Naseva, E.; Miteva, D.G.; Velikova, T. Artificial intelligence as a tool in drug discovery and development.
World J. Exp. Med. 2024,14, 96042. [CrossRef] [PubMed]
223.
Amin, M.; Martínez-Heras, E.; Ontaneda, D.; Carrasco, F.P. Artificial Intelligence and Multiple Sclerosis. Curr. Neurol. Neurosci.
Rep. 2024,24, 233. [CrossRef]
224.
Kurniawan, I.T.; Guitart-Masip, M.; Dolan, R.J. Dopamine and Effort-Based Decision Making. Front. Neurosci. 2011,5, 81.
[CrossRef]
225.
Liu, X.; Zhang, J.; Hou, Z.; Yang, Y.I.; Gao, Y.Q. From predicting to decision making: Reinforcement learning in biomedicine.
WIREs Comput. Mol. Sci. 2024,14, e1723. [CrossRef]
226.
Jankowsky, K.; Krakau, L.; Schroeders, U.; Zwerenz, R.; Beutel, M.E. Predicting treatment response using machine learning: A
registered report. Br. J. Clin. Psychol. 2024,63, 137–155. [CrossRef]
227.
Wu, Y.; Mao, K.; Dennett, L.; Zhang, Y.; Chen, J. Systematic review of machine learning in PTSD studies for automated diagnosis
evaluation. Npj Ment. Health Res. 2023,2, 16. [CrossRef]
228.
Winchester, L.M.; Harshfield, E.L.; Shi, L.; Badhwar, A.; Khleifat, A.A.; Clarke, N.; Dehsarvi, A.; Lengyel, I.; Lourida, I.; Madan,
C.R.; et al. Artificial intelligence for biomarker discovery in Alzheimer’s disease and dementia. Alzheimers Dement. J. Alzheimers
Assoc. 2023,19, 5860–5871. [CrossRef]
229.
Venugopalan, J.; Tong, L.; Hassanzadeh, H.R.; Wang, M.D. Multimodal deep learning models for early detection of Alzheimer’s
disease stage. Sci. Rep. 2021,11, 3254. [CrossRef]
230.
Ul Rehman, S.; Tarek, N.; Magdy, C.; Kamel, M.; Abdelhalim, M.; Melek, A.; Mahmoud, L.N.; Sadek, I. AI-based tool for early
detection of Alzheimer’s disease. Heliyon 2024,10, e29375. [CrossRef]
231.
Doherty, T.; Yao, Z.; Khleifat, A.A.; Tantiangco, H.; Tamburin, S.; Albertyn, C.; Thakur, L.; Llewellyn, D.J.; Oxtoby, N.P.;
Lourida, I.; et al. Artificial intelligence for dementia drug discovery and trials optimization. Alzheimers Dement. 2023,19,
5922–5933. [CrossRef] [PubMed]
232.
Angelucci, F.; Ai, A.R.; Piendel, L.; Cerman, J.; Hort, J. Integrating AI in fighting advancing Alzheimer: Diagnosis, prevention,
treatment, monitoring, mechanisms, and clinical trials. Curr. Opin. Struct. Biol. 2024,87, 102857. [CrossRef] [PubMed]
233.
Yamakawa, T.; Miyajima, M.; Fujiwara, K.; Kano, M.; Suzuki, Y.; Watanabe, Y.; Watanabe, S.; Hoshida, T.; Inaji, M.; Maehara, T.
Wearable Epileptic Seizure Prediction System with Machine-Learning-Based Anomaly Detection of Heart Rate Variability. Sensors
2020,20, 3987. [CrossRef]
234.
Yuan, S.; Yan, K.; Wang, S.; Liu, J.-X.; Wang, J. EEG-Based Seizure Prediction Using Hybrid DenseNet–ViT Network with Attention
Fusion. Brain Sci. 2024,14, 839. [CrossRef]
235.
An, S.; Kang, C.; Lee, H.W. Artificial Intelligence and Computational Approaches for Epilepsy. J. Epilepsy Res. 2020,10, 8.
[CrossRef]
236. Giansanti, D. An Umbrella Review of the Fusion of fMRI and AI in Autism. Diagnostics 2023,13, 3552. [CrossRef]
237.
Rasul, R.A.; Saha, P.; Bala, D.; Karim, S.M.R.U.; Abdullah, M.I.; Saha, B. An evaluation of machine learning approaches for early
diagnosis of autism spectrum disorder. Healthc. Anal. 2024,5, 100293. [CrossRef]
238.
Murdaugh, D.L.; Nadendla, K.D.; Kana, R.K. Differential role of temporoparietal junction and medial prefrontal cortex in causal
inference in autism: An independent component analysis. Neurosci. Lett. 2014,568, 50–55. [CrossRef]
239.
Zafar, F.; Alam, L.F.; Vivas, R.R.; Wang, J.; Whei, S.J.; Mehmood, S.; Sadeghzadegan, A.; Lakkimsetti, M.; Nazir, Z. The Role
of Artificial Intelligence in Identifying Depression and Anxiety: A Comprehensive Literature Review. Cureus 2024,16, e56472.
[CrossRef]
240.
Xia, Z.; Fan, Y.; Li, K.; Wang, Y.; Huang, L.; Zhou, F. DepressionGraph: A Two-Channel Graph Neural Network for the Diagnosis
of Major Depressive Disorders Using rs-fMRI. Electronics 2023,12, 5040. [CrossRef]
241.
Taliaz, D.; Spinrad, A.; Barzilay, R.; Barnett-Itzhaki, Z.; Averbuch, D.; Teltsh, O.; Schurr, R.; Darki-Morag, S.; Lerer, B. Optimizing
prediction of response to antidepressant medications using machine learning and integrated genetic, clinical, and demographic
data. Transl. Psychiatry 2021,11, 381. [CrossRef] [PubMed]
J. Clin. Med. 2025,14, 550 43 of 45
242.
Olawade, D.B.; Wada, O.Z.; Odetayo, A.; David-Olawade, A.C.; Asaolu, F.; Eberhardt, J. Enhancing mental health with Artificial
Intelligence: Current trends and future prospects. J. Med. Surg. Public Health 2024,3, 100099. [CrossRef]
243.
Abedi, V.; Khan, A.; Chaudhary, D.; Misra, D.; Avula, V.; Mathrawala, D.; Kraus, C.; Marshall, K.A.; Chaudhary, N.; Li, X.; et al.
Using artificial intelligence for improving stroke diagnosis in emergency departments: A practical framework. Ther. Adv. Neurol.
Disord. 2020,13, 1756286420938962. [CrossRef] [PubMed]
244.
Koska, I.O.; Selver, A. Artificial Intelligence in Stroke Imaging: A Comprehensive Review. Eurasian J. Med. 2023,55, S91.
[CrossRef]
245. Petrella, R.J. The AI Future of Emergency Medicine. Ann. Emerg. Med. 2024,84, 139–153. [CrossRef]
246.
Wu, P.; Cao, B.; Liang, Z.; Wu, M. The advantages of artificial intelligence-based gait assessment in detecting, predicting, and
managing Parkinson’s disease. Front. Aging Neurosci. 2023,15, 1191378. [CrossRef]
247.
Kamran, I.; Naz, S.; Razzak, I.; Imran, M. Handwriting dynamics assessment using deep neural network for early identification of
Parkinson’s disease. Future Gener. Comput. Syst. 2021,117, 234–244. [CrossRef]
248.
Reddy, A.; Reddy, R.P.; Roghani, A.K.; Garcia, R.I.; Khemka, S.; Pattoor, V.; Jacob, M.; Reddy, P.H.; Sehar, U. Artificial intelligence
in Parkinson’s disease: Early detection and diagnostic advancements. Ageing Res. Rev. 2024,99, 102410. [CrossRef]
249.
Bounsall, K.; Milne-Ives, M.; Hall, A.; Carroll, C.; Meinert, E. Artificial Intelligence Applications for Assessment, Monitoring, and
Management of Parkinson Disease Symptoms: Protocol for a Systematic Review. JMIR Res. Protoc. 2023,12, e46581. [CrossRef]
250.
AbuAlrob, M.A.; Mesraoua, B. Harnessing artificial intelligence for the diagnosis and treatment of neurological emergencies: A
comprehensive review of recent advances and future directions. Front. Neurol. 2024,15, 1485799. [CrossRef]
251.
Wojtara, M.; Rana, E.; Rahman, T.; Khanna, P.; Singh, H. Artificial intelligence in rare disease diagnosis and treatment. Clin. Transl.
Sci. 2023,16, 2106. [CrossRef] [PubMed]
252.
He, D.; Wang, R.; Xu, Z.; Wang, J.; Song, P.; Wang, H.; Su, J. The use of artificial intelligence in the treatment of rare diseases: A
scoping review. Intractable Rare Dis. Res. 2024,13, 12. [CrossRef] [PubMed]
253.
van de Leur, R.R.; Boonstra, M.J.; Bagheri, A.; Roudijk, R.W.; Sammani, A.; Taha, K.; Doevendans, P.A.; van der Harst, P.; van
Dam, P.M.; Hassink, R.J.; et al. Big Data and Artificial Intelligence: Opportunities and Threats in Electrophysiology. Arrhythmia
Electrophysiol. Rev. 2020,9, 146. [CrossRef] [PubMed]
254.
Stripelis, D.; Gupta, U.; Saleem, H.; Dhinagar, N.; Ghai, T.; Anastasiou, C.; Sánchez, R.; Steeg, G.V.; Ravi, S.; Naveed, M.; et al. A
federated learning architecture for secure and private neuroimaging analysis. Patterns 2024,5, 101031. [CrossRef]
255.
Corcuera Bárcena, J.L.; Ducange, P.; Marcelloni, F.; Renda, A. Increasing trust in AI through privacy preservation and model
explainability: Federated Learning of Fuzzy Regression Trees. Inf. Fusion 2025,113, 102598. [CrossRef]
256.
Markiewicz, C.J.; Gorgolewski, K.J.; Feingold, F.; Blair, R.; Halchenko, Y.O.; Miller, E.; Hardcastle, N.; Wexler, J.; Esteban, O.;
Goncavles, M.; et al. The OpenNeuro resource for sharing of neuroscience data. eLife 2021,10, e71774. [CrossRef]
257.
Pati, S.; Kumar, S.; Varma, A.; Edwards, B.; Lu, C.; Qu, L.; Wang, J.J.; Lakshminarayanan, A.; Wang, S.; Sheller, M.J.; et al. Privacy
preservation for federated learning in health care. Patterns 2024,5, 100974. [CrossRef]
258.
Kiseleva, A.; Kotzinos, D.; Hert, P.D. Transparency of AI in Healthcare as a Multilayered System of Accountabilities: Between
Legal Requirements and Technical Limitations. Front. Artif. Intell. 2022,5, 879603. [CrossRef]
259.
Xu, H.; Shuttleworth, K.M.J. Medical artificial intelligence and the black box problem: A view based on the ethical principle of
“do no harm”. Intell. Med. 2024,4, 52–57. [CrossRef]
260.
Sadeghi, Z.; Alizadehsani, R.; Cifci, M.A.; Kausar, S.; Rehman, R.; Mahanta, P.; Bora, P.K.; Almasri, A.; Alkhawaldeh, R.S.; Hussain,
S.; et al. A review of Explainable Artificial Intelligence in healthcare. Comput. Electr. Eng. 2024,118, 109370. [CrossRef]
261.
Mienye, I.D.; Obaido, G.; Jere, N.; Mienye, E.; Aruleba, K.; Emmanuel, I.D.; Ogbuokiri, B. A survey of explainable artificial
intelligence in healthcare: Concepts, applications, and challenges. Inform. Med. Unlocked 2024,51, 101587. [CrossRef]
262.
Tilala, M.H.; Chenchala, P.K.; Choppadandi, A.; Kaur, J.; Naguri, S.; Saoji, R.; Devaguptapu, B. Ethical Considerations in the Use
of Artificial Intelligence and Machine Learning in Health Care: A Comprehensive Review. Cureus 2024,16, e62443. [CrossRef]
263.
White, T.; Blok, E.; Calhoun, V.D. Data sharing and privacy issues in neuroimaging research: Opportunities, obstacles, challenges,
and monsters under the bed. Hum. Brain Mapp. 2020,43, 278. [CrossRef]
264. Salles, A.; Farisco, M. Neuroethics and AI ethics: A proposal for collaboration. BMC Neurosci. 2024,25, 41. [CrossRef]
265.
Ahmed, M.I.; Spooner, B.; Isherwood, J.; Lane, M.; Orrock, E.; Dennison, A. A Systematic Review of the Barriers to the
Implementation of Artificial Intelligence in Healthcare. Cureus 2023,15, e46454. [CrossRef]
266.
Li, A.; Guan, Y.; Gong, H.; Luo, Q. Challenges of Processing and Analyzing Big Data in Mesoscopic Whole-brain Imaging. Genom.
Proteom. Bioinform. 2019,17, 337. [CrossRef]
267.
Prangon, N.F.; Wu, J. AI and Computing Horizons: Cloud and Edge in the Modern Era. J. Sens. Actuator Netw. 2024,13, 44.
[CrossRef]
268.
Asan, O.; Bayrak, A.E.; Choudhury, A. Artificial Intelligence and Human Trust in Healthcare: Focus on Clinicians. J. Med. Internet
Res. 2020,22, e15154. [CrossRef]
J. Clin. Med. 2025,14, 550 44 of 45
269.
Hassan, M.; Kushniruk, A.; Borycki, E. Barriers to and Facilitators of Artificial Intelligence Adoption in Health Care: Scoping
Review. JMIR Hum. Factors 2024,11, e48633. [CrossRef]
270.
Palaniappan, K.; Lin, E.Y.T.; Vogel, S. Global Regulatory Frameworks for the Use of Artificial Intelligence (AI) in the Healthcare
Services Sector. Healthcare 2024,12, 562. [CrossRef]
271.
Yousif, M.; Asghar, S.; Akbar, J.; Masood, I.; Arshad, M.R.; Naeem, J.; Azam, A.; Iqbal, Z. Exploring the perspectives of healthcare
professionals regarding artificial intelligence; acceptance and challenges. BMC Health Serv. Res. 2024,24, 1200. [CrossRef]
[PubMed]
272.
Krichmar, J.L.; Olds, J.L.; Sanchez-Andres, J.V.; Tang, H. Editorial: Explainable Artificial Intelligence and Neuroscience: Cross-
Disciplinary Perspectives. Front. Neurorobot. 2021,15, 731733. [CrossRef] [PubMed]
273.
Nagendran, M.; Festor, P.; Komorowski, M.; Gordon, A.C.; Faisal, A.A. Quantifying the impact of AI recommendations with
explanations on prescription decision making. Npj Digit. Med. 2023,6, 206. [CrossRef] [PubMed]
274.
Arora, A.; Arora, A. Generative adversarial networks and synthetic patient data: Current challenges and future perspectives.
Future Healthc. J. 2022,9, 190. [CrossRef] [PubMed]
275.
Michelutti, L.; Tel, A.; Zeppieri, M.; Ius, T.; Agosti, E.; Sembronio, S.; Robiony, M. Generative Adversarial Networks (GANs) in
the Field of Head and Neck Surgery: Current Evidence and Prospects for the Future—A Systematic Review. J. Clin. Med. 2024,
13, 3556. [CrossRef]
276.
Raikar, A.S.; Andrew, J.; Dessai, P.P.; Prabhu, S.M.; Jathar, S.; Prabhu, A.; Naik, M.B.; Raikar, G.V.S. Neuromorphic computing for
modeling neurological and psychiatric disorders: Implications for drug development. Artif. Intell. Rev. 2024,57, 318. [CrossRef]
277.
Currie, G.M. The emerging role of artificial intelligence and digital twins in pre-clinical molecular imaging. Nucl. Med. Biol. 2023,
120–121, 108337. [CrossRef]
278. Alam El Din, D.-M.; Shin, J.; Lysinger, A.; Roos, M.J.; Johnson, E.C.; Shafer, T.J.; Hartung, T.; Smirnova, L. Organoid intelligence
for developmental neurotoxicity testing. Front. Cell. Neurosci. 2024,18, 1480845. [CrossRef]
279.
Pun, F.W.; Ozerov, I.V.; Zhavoronkov, A. AI-powered therapeutic target discovery. Trends Pharmacol. Sci. 2023,44, 561–572.
[CrossRef]
280.
Bertacchini, F.; Demarco, F.; Scuro, C.; Pantano, P.; Bilotta, E. A social robot connected with chatGPT to improve cognitive
functioning in ASD subjects. Front. Psychol. 2023,14, 1232177. [CrossRef]
281.
Chen, Z.; Yadollahpour, A. A new era in cognitive neuroscience: The tidal wave of artificial intelligence (AI). BMC Neurosci. 2024,
25, 23. [CrossRef] [PubMed]
282.
Wang, S.; Lim, W.M.; Cheah, J.-H.; Lim, X.-J. Working with robots: Trends and future directions. Technol. Forecast. Soc. Chang.
2025,212, 123648. [CrossRef]
283.
Jiang, T.; Sun, Z.; Fu, S.; Lv, Y. Human-AI interaction research agenda: A user-centered perspective. Data Inf. Manag. 2024,8,
100078. [CrossRef]
284.
Zhu, B.; Shin, U.; Shoaran, M. Closed-Loop Neural Prostheses with On-Chip Intelligence: A Review and A Low-Latency Machine
Learning Model for Brain State Detection. IEEE Trans. Biomed. Circuits Syst. 2021,15, 877. [CrossRef]
285.
Oliveira, A.M.; Coelho, L.; Carvalho, E.; Ferreira-Pinto, M.J.; Vaz, R.; Aguiar, P. Machine learning for adaptive deep brain
stimulation in Parkinson’s disease: Closing the loop. J. Neurol. 2023,270, 5313. [CrossRef]
286.
Bobo, W.V.; Van Ommeren, B.; Athreya, A.P. Machine learning, pharmacogenomics, and clinical psychiatry: Predicting antide-
pressant response in patients with major depressive disorder. Expert Rev. Clin. Pharmacol. 2022,15, 927–944. [CrossRef]
287.
Moggia, D.; Lutz, W.; Brakemeier, E.-L.; Bickman, L. Treatment Personalization and Precision Mental Health Care: Where are we
and where do we want to go? Adm. Policy Ment. Health 2024,51, 611. [CrossRef]
288.
Sokaˇc, M.; Mrši´c, L.; Balkovi´c, M.; Brkljaˇci´c, M. Bridging Artificial Intelligence and Neurological Signals (BRAINS): A Novel
Framework for Electroencephalogram-Based Image Generation. Information 2024,15, 405. [CrossRef]
289.
Apostolova, L.G.; Aisen, P.; Eloyan, A.; Fagan, A.; Fargo, K.N.; Foroud, T.; Gatsonis, C.; Grinberg, L.T.; Clifford R Jack, J.;
Kramer, J.; et al. The Longitudinal Early-onset Alzheimer’s Disease Study (LEADS): Framework and methodology. Alzheimers
Dement. 2021,17, 2043. [CrossRef]
290.
Amiri, M.; Nazari, S.; Jafari, A.H.; Makkiabadi, B. A new full closed-loop brain-machine interface approach based on neural
activity: A study based on modeling and experimental studies. Heliyon 2023,9, e13766. [CrossRef]
291.
Aderinto, N.; AbdulBasit, M.O.; Olatunji, G.; Adejumo, T. Exploring the transformative influence of neuroplasticity on stroke
rehabilitation: A narrative review of current evidence. Ann. Med. Surg. 2023,85, 4425. [CrossRef] [PubMed]
292.
Meribout, M.; Abule Takele, N.; Derege, O.; Rifiki, N.; El Khalil, M.; Tiwari, V.; Zhong, J. Tactile sensors: A review. Measurement
2024,238, 115332. [CrossRef]
293.
Sarker, I.H. Machine Learning: Algorithms, Real-World Applications and Research Directions. SN Comput. Sci. 2021,2, 160.
[CrossRef] [PubMed]
294.
Drigas, A.; Sideraki, A. Brain Neuroplasticity Leveraging Virtual Reality and Brain–Computer Interface Technologies. Sensors
2024,24, 5725. [CrossRef]
J. Clin. Med. 2025,14, 550 45 of 45
295.
Naseer, F.; Khan, M.N.; Tahir, M.; Addas, A.; Aejaz, S.M.H. Integrating deep learning techniques for personalized learning
pathways in higher education. Heliyon 2024,10, e32628. [CrossRef]
296.
Beauchemin, N.; Charland, P.; Karran, A.; Boasen, J.; Tadson, B.; Sénécal, S.; Léger, P.-M. Enhancing learning experiences:
EEG-based passive BCI system adapts learning speed to cognitive load in real-time, with motivation as catalyst. Front. Hum.
Neurosci. 2024,18, 1416683. [CrossRef]
297. Holmes, W.; Tuomi, I. State of the art and practice in AI in education. Eur. J. Educ. 2022,57, 542–570. [CrossRef]
298.
Safdar, N.M.; Banja, J.D.; Meltzer, C.C. Ethical considerations in artificial intelligence. Eur. J. Radiol. 2020,122, 108768. [CrossRef]
299.
Cachat-Rosset, G.; Klarsfeld, A. Diversity, Equity, and Inclusion in Artificial Intelligence: An Evaluation of Guidelines. Appl. Artif.
Intell. 2023,37, 2176618. [CrossRef]
300.
Ferrara, E. Fairness and Bias in Artificial Intelligence: A Brief Survey of Sources, Impacts, and Mitigation Strategies. Sci 2024,6, 3.
[CrossRef]
301.
Kusters, R.; Misevic, D.; Berry, H.; Cully, A.; Le Cunff, Y.; Dandoy, L.; Díaz-Rodríguez, N.; Ficher, M.; Grizou, J.; Othmani, A.; et al.
Interdisciplinary Research in Artificial Intelligence: Challenges and Opportunities. Front. Big Data 2020,3, 577974. [CrossRef]
[PubMed]
302.
Roche, C.; Wall, P.J.; Lewis, D. Ethics and diversity in artificial intelligence policies, strategies and initiatives. AI Ethics 2023,3,
1095–1115. [CrossRef] [PubMed]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.
Available via license: CC BY 4.0
Content may be subject to copyright.
