Simon B. Eickhoff’s research while affiliated with University of Bonn and other places
What is this page?
This page lists works of an author who doesn't have a ResearchGate profile or hasn't added the works to their profile yet. It is automatically generated from public (personal) data to further our legitimate goal of comprehensive and accurate scientific recordkeeping. If you are this author and want this page removed, please let us know.
Machine learning (ML) provides powerful tools for predictive modeling. ML’s popularity stems from the promise of sample-level prediction with applications across a variety of fields from physics and marketing to healthcare. However, if not properly implemented and evaluated, ML pipelines may contain leakage typically resulting in overoptimistic performance estimates and failure to generalize to new data. This can have severe negative financial and societal implications. Our aim is to expand understanding associated with causes leading to leakage when designing, implementing, and evaluating ML pipelines. Illustrated by concrete examples, we provide a comprehensive overview and discussion of various types of leakage that may arise in ML pipelines.
Insomnia affects a substantial proportion of the population and frequently co-occurs with mental illnesses including depression and anxiety. However, the neurobiological correlates of these disorders remain unclear. Here we review magnetic resonance imaging (MRI) studies assessing structural and functional brain associations with depressive and anxiety symptoms in insomnia disorder (ID; n = 38), insomnia symptoms in depressive and anxiety disorders (n = 14), and these symptoms in the general populations (n = 3). The studies on insomnia disorder consistently showed overlapping (salience network: insula and anterior cingulate cortex) and differential MRI correlation patterns between depressive (thalamus, orbitofrontal cortex and its associated functional connectivity) and anxiety (functional connectivity associated with default mode network) symptoms. The insula was also consistently identified as indicating the severity of insomnia symptoms in depressive disorder. In contrast, findings for other regions related to insomnia symptoms in both depressive and anxiety disorders were generally inconsistent across studies, partly due to variations in methods and patient cohorts. In the general population, brain regions in the default mode network provided a functional link between insomnia and depressive symptoms. These findings underscore both the shared and distinct neural correlates among depression, anxiety, and insomnia, providing potential avenues for the clinical management of these conditions.
Recent neuroimaging studies demonstrated that task-evoked functional connectivity (FC) may better predict individual traits than resting-state FC. However, the prediction properties of task-evoked effective connectivity (EC) remain unexplored. We investigated this by predicting individual reaction time (RT) performance in the stimulus-response compatibility task and age, using intrinsic EC (I-EC; calculated at baseline) and task-modulated EC (M-EC; induced by experimental conditions) with dynamic causal modeling (DCM) across various data processing conditions, including different general linear model (GLM) designs, Bayesian model reduction, and different cross-validation schemes and prediction models. We report evident differences in predicting RT and age between I-EC and M-EC, as well as between event-related and block-based GLM and DCM designs. M-EC outperformed both I-EC and task-evoked FC in RT prediction, while all types of connectivity performed similarly for age. Event-related GLM and DCM designs performed better than block-based designs. Our findings suggest that task-evoked I-EC and M-EC may capture different phenotypic attributes, with performance influenced by data processing and modeling choices, particularly the GLM-DCM design. This evaluation of methods for behavior prediction from brain EC may contribute to a meta-scientific understanding of how data processing and modeling frameworks influence neuroimaging-based predictions, offering insights for improving their robustness and efficacy.
Determining the brain specializations unique to humans requires directly comparative anatomical information from other primates, especially our closest relatives. Human ( Homo sapiens ) (m/f), chimpanzee ( Pan troglodytes ) (f), and rhesus macaque ( Macaca mulatta ) (m/f) white matter atlases were used to create connectivity blueprints, i.e., descriptions of the cortical grey matter in terms of the connectivity with homologous white matter tracts. This allowed a quantitative comparative of cortical organization across the species. We identified human-unique connectivity profiles concentrated in temporal and parietal cortices, and hominid-unique organization in prefrontal cortex. Functional decoding revealed human-unique hotspots correlated with language processing and social cognition. Overall, our results counter models that assign primacy to prefrontal cortex for human uniqueness.
Significance statement Understanding what makes the human brain unique requires direct comparisons with other primates, particularly our closest relatives. Using connectivity blueprints, we compared to cortical organization of the human to that of the macaque and, for the first time, the chimpanzee. This approach revealed human-specific connectivity patterns in the temporal and parietal lobes, regions linked to language and social cognition. These findings challenge traditional views that prioritize the prefrontal cortex in defining human cognitive uniqueness, emphasizing instead the importance of temporal and parietal cortical evolution in shaping our species’ abilities.
The rapid developments in the field of artificial intelligence (AI) are opening up numerous new diagnostic and therapeutic options in pediatric and adolescent medicine. At the same time, however, this development also raises a variety of questions. In this article, we highlight the factors that are important for the acceptance of AI in pediatrics and adolescent medicine and outline some ethical and legal challenges. An interdisciplinary perspective and a careful handling of concerns and fears is essential to meet these challenges and to take advantage of the opportunities that AI offers.
Activation Likelihood Estimation (ALE) employs voxel- or cluster-level family-wise error (vFWE or cFWE) correction or threshold-free cluster enhancement (TFCE) to counter false positives due to multiple comparisons. These corrections utilize Monte-Carlo simulations to approximate a null distribution of spatial convergence, which allows for the determination of a corrected significance threshold. The simulations may take many hours depending on the dataset and the hardware used to run the computations. In this study, we aimed to replace the time-consuming Monte-Carlo simulation procedure with an instantaneous machine-learning prediction based on features of the meta-analysis dataset. These features were created from the number of experiments in the dataset, the number of subjects per experiment, and the number of foci reported per experiment. We simulated 68,100 training datasets, containing between 10 and 150 experiments and computed the vFWE, cFWE, and TFCE significance thresholds. We then used this data to train one XGBoost regression model for each thresholding technique. Lastly, we validated the performance of the three models using 11 independent real-life datasets (21 contrasts) from previously published ALE meta-analyses. The vFWE model reached near-perfect prediction levels (R² = 0.996), while the TFCE and cFWE models achieved very good prediction accuracies of R² = 0.951 and R² = 0.938, respectively. This means that, on average, the difference between predicted and standard (monte-carlo based) cFWE thresholds was less than two voxels. Given that our model predicts significance thresholds in ALE meta-analyses with very high accuracy, we advocate our efficient prediction approach as a replacement for the currently used Monte-Carlo simulations in future ALE analyses. This will save hours of computation time and reduce energy consumption. Furthermore, the reduced compute time allows for easier implementation of multi-analysis set-ups like leave-one-out sensitivity analysis or subsampling.
The integration of artificial intelligence (AI) and machine learning (ML) has shown potential for various applications in the medical field, particularly for diagnosing and managing chronic diseases among children and adolescents. This systematic review aims to comprehensively analyze and synthesize research on the use of AI for monitoring, guiding, and assisting pediatric patients with chronic diseases. Five major electronic databases were searched (Medline, Scopus, PsycINFO, ACM, Web of Science), along with manual searches of gray literature, personal archives, and reference lists of relevant papers. All original studies as well as conference abstracts and proceedings, focusing on AI applications for pediatric chronic disease care were included. Thirty-one studies met the inclusion criteria. We extracted AI method used, study design, population, intervention, and main results. Two researchers independently extracted data and resolved discrepancies through discussion. AI applications are diverse, encompassing, e.g., disease classification, outcome prediction, or decision support. AI generally performed well, though most models were tested on retrospective data. AI-based tools have shown promise in mental health analysis, e.g., by using speech sampling or social media data to predict therapy outcomes for various chronic conditions.
Conclusions: While AI holds potential in pediatric chronic disease care, most reviewed studies are small-scale research projects. Prospective clinical implementations are needed to validate its effectiveness in real-world scenarios. Ethical considerations, cultural influences, and stakeholder attitudes should be integrated into future research. What is known:
• Artificial Intelligence (AI) will play a more dominant role in medicine and healthcare in the future and many applications are already being developed.
What is new:
• Our review provides an overview on how AI-driven systems might be able to support children and adolescents with chronic illnesses.
• While many applications are being researched, few have been tested on real-world, prospective, clinical data.
Cognitive impairment is a central characteristic of schizophrenia. Executive functioning (EF) impairments are often seen in mental disorders, particularly schizophrenia, where they relate to adverse outcomes. As a heterogeneous construct, how specifically each dimension of EF to characterize the diagnostic and prognostic aspects of schizophrenia remains opaque. We used classification models with a stacking approach on systematically measured EFs to discriminate 195 patients with schizophrenia from healthy individuals. Baseline EF measurements were moreover employed to predict symptomatically remitted or non-remitted prognostic subgroups. EF feature importance was determined at the group-level and the ensuing individual importance scores were associated with four symptom dimensions. EF assessments of inhibitory control (interference and response inhibitions), followed by working memory, evidently predicted schizophrenia diagnosis (area under the curve [AUC]=0.87) and remission status (AUC=0.81). The models highlighted the importance of interference inhibition or working memory updating in accurately identifying individuals with schizophrenia or those in remission. These identified patients had high-level negative symptoms at baseline and those who remitted showed milder cognitive symptoms at follow-up, without differences in baseline EF or symptom severity compared to non-remitted patients. Our work indicates that impairments in specific EF dimensions in schizophrenia are differentially linked to individual symptom-load and prognostic outcomes. Thus, assessments and models based on EF may be a promising tool that can aid in the clinical evaluation of this disorder.
... At the same time, insomnia and poor sleep quality themselves function as transdiagnostic risk factors for developing severe mental health conditions, including depression and anxiety [79,82,83]. These effects are potentially mediated by shared disruptions in neurobiological systems that regulate both sleep and emotion [84]. ...
... blue, orange and green rows in Fig. 1B-top) and homologous brain regions can be identified on the basis of tractography fingerprint similarity (Mars et al. 2018a(Mars et al. , 2021. A number of studies have proposed tractography protocols for reconstructing corresponding WM bundles whose central cores are known to exist in both humans and NHPs, for instance, in macaques and humans (Warrington et al. 2020;Assimopoulos et al. 2024), chimpanzees and humans (Bryant et al. , 2025 or great apes and humans (Roumazeilles et al. 2020). Tractography fingerprints can then be estimated for all regions with respect to this same set of WM bundles, providing a common space for performing comparative comparisons (Mars et al. 2018b). ...
... For instance, an AI-powered interactive toy reduced perceived postoperative pain by 25% and enhanced coping strategies by 20% compared to conventional methods [18]. Likewise, AI-assisted relaxation techniques significantly decreased preoperative anxiety (p < 0.05) and led to notable improvements in caregiverreported quality of life [19]. ...
... Existing evidence suggested several shared underlying mechanisms for the insomnia-depressive symptoms link, including shared genetic factors, hyperactivity of the arousal system, disruptions in synaptic plasticity, alterations in the biological timekeeping system, and shared macroscale neuroanatomical abnormalities between insomnia and depression [39,40,41,42,43]. We observed a trend of increasing multivariate associations between insomnia and depressive symptoms after psychotherapy. ...
... Functional atlases are primarily designed to reflect connectivity patterns, which may not always correspond directly to the brain's structural organization. Despite these concerns, we selected the Schaefer atlas based on its established utility in previous studies Kuchenhoff et al., 2024;Luppi et al., 2024;Serio et al., 2024;Valk et al., 2020;L. Zhang et al., 2025), where it demonstrated robust performance in extracting biologically relevant features even from structural data. ...
... But I cannot adopt such a use, since in my view-Bejarano (2022)-, self-control is not necessarily moral). 29 Could Bryant et al. (2024) reinforce that claim? They state: "Our findings support a two-step evolutionary process, in which changes in prefrontal cortex organization emerge prior to changes in temporal areas". ...
... Here, one important avenue of research is targeting the neurocognitive pathway of mentalizing by means of NIBS in order to facilitate ASD indi-viduals´ inferences regarding other peoples' mental states [20]. While previous research has provided proof-ofprinciple evidence that repetitive transcranial magnetic stimulation (rTMS) of the mentalizing pathway can modulate social abilities in patients with ASD, the results of NIBS studies with ASD individuals have not been consistent [6,[21][22][23][24][25][26][27][28][29]. Still, the majority of the few existing studies investigated the effects of rTMS over the dorsolater prefrontal cortex (dlPFC) attempting to induce NIBS effects on executive functions, but showed no improvement in social deficits [21][22][23][24][25]. ...
... While not specific to any neurodegenerative disorders, the concept of brain age can assist the understanding of disease interaction with the normal aging process. For instance, pathological accumulation of amyloid and tau proteins (both implicated in the progression of AD and other neurodegenerative disorders) are significantly correlated to BAG in patients with mild cognitive impairment and healthy controls (Doering et al., 2024). ...
... The dlPFC is known for its involvement in cognitive control such as topdown adjustment of response inhibition (31), and is also known as a regulator of adolescent risky decision making (32). Although not examining functional connectivity directly, recent meta-analyses also indicate that dACC and dlPFC coactivation is crucial in risk-taking behaviors (36,37). Additionally, a systematic review on neural correlates of risk-taking in substance-related behaviors found that dlPFC activation decreases, whereas dACC activity increases, during risk-taking (38). ...
... Recently, ML models have started being increasingly applied to predict outcomes in child development -a notoriously complex and multi-variate domain of inquiry. In a recent systematic review, Reinhart et al. (2024) highlight the promise of artificial intelligence in this area, showing how ML techniques can enhance our capacity to monitor typical development, detect early signs of delay, and support early diagnosis of developmental disorders. So far, ML models have been used in the early prediction of cognitive and language disorders (e.g., Borovsky et al., 2021;Demirci et al., 2024;Lammert et al., 2025;Shenkin, 2023;Toki et al., 2024), as well as to identify conditions such as Autism Spectrum Disorder (ASD), ADHD, or other neurodevelopmental disorders (e.g., Ayub et al., 2024;Ortega León et al., 2025;Wei et al., 2023). ...
![A schematic representation of the cross validation (CV) scheme: Here, we illustrate a single repeat of a \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:k$$\end{document}-fold CV with three folds (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:k=3$$\end{document}) with the third fold being used as the test data](https://www.researchgate.net/publication/392203616/figure/fig1/AS:11431281475886505@1748573197103/A-schematic-representation-of-the-cross-validation-CV-scheme-Here-we-illustrate-a_Q320.jpg)
