Dustin Scheinost’s research while affiliated with Yale University 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.

Publications (372)


Neural Variability and Cognitive Control in Individuals With Opioid Use Disorder
  • Article

January 2025

·

20 Reads

JAMA Network Open

Jean Ye

·

Saloni Mehta

·

·

[...]

·

Dustin Scheinost

Importance Opioid use disorder (OUD) impacts millions of people worldwide. Prior studies investigating its underpinning neural mechanisms have not often considered how brain signals evolve over time, so it remains unclear whether brain dynamics are altered in OUD and have subsequent behavioral implications. Objective To characterize brain dynamic alterations and their association with cognitive control in individuals with OUD. Design, Setting, and Participants This case-control study collected functional magnetic resonance imaging (fMRI) data from individuals with OUD and healthy control (HC) participants. The study was performed at an academic research center and an outpatient clinic from August 2019 to May 2024. Exposure Individuals with OUD were all recently stabilized on medications for OUD (<24 weeks). Main Outcomes and Measures Recurring brain states supporting different cognitive processes were first identified in an independent sample with 390 participants. A multivariate computational framework extended these brain states to the current dataset to assess their moment-to-moment engagement within each individual. Resting-state and naturalistic fMRI investigated whether brain dynamic alterations were consistently observed in OUD. Using a drug cue paradigm in participants with OUD, the association between cognitive control and brain dynamics during exposure to opioid-related information was studied. Variations in continuous brain state engagement (ie, state engagement variability [SEV]) were extracted during resting-state, naturalistic, and drug-cue paradigms. Stroop assessed cognitive control. Results Overall, 99 HC participants (54 [54.5%] female; mean [SD] age, 31.71 [12.16] years) and 76 individuals with OUD (31 [40.8%] female; mean [SD] age, 39.37 [10.47] years) were included. Compared with HC participants, individuals with OUD demonstrated consistent SEV alterations during resting-state (99 HC participants; 71 individuals with OUD; F 4,161 = 6.83; P < .001) and naturalistic (96 HC participants; 76 individuals with OUD; F 4,163 = 9.93; P < .001) fMRI. Decreased cognitive control was associated with lower SEV during the rest period of a drug cue paradigm among 70 participants with OUD. For example, lower incongruent accuracy scores were associated with decreased transition SEV (ρ 58 = 0.34; P = .008). Conclusions and Relevance In this case-control study of brain dynamics in OUD, individuals with OUD experienced greater difficulty in effectively engaging various brain states to meet changing demands. Decreased cognitive control during the rest period of a drug cue paradigm suggests that these individuals had an impaired ability to disengage from opioid-related information. The current study introduces novel information that may serve as groundwork to strengthen cognitive control and reduce opioid-related preoccupation in OUD.


What is the best brain state to predict autistic traits?
  • Preprint
  • File available

January 2025

·

42 Reads

Autism is a heterogeneous condition, and functional magnetic resonance imaging-based studies have advanced understanding of neurobiological correlates of autistic features. Nevertheless, little work has focused on the optimal brain states to reveal brain-phenotype relationships. In addition, there is a need to better understand the relevance of attentional abilities in mediating autistic features. Using connectome-based predictive modelling, we interrogate three datasets to determine scanning conditions that can boost prediction of clinically relevant phenotypes and assess generalizability. In dataset one, a sample of youth with autism and neurotypical participants, we find that a sustained attention task (the gradual onset continuous performance task) results in high prediction performance of autistic traits compared to a free-viewing social attention task and a resting-state condition. In dataset two, we observe the predictive network model of autistic traits generated from the sustained attention task generalizes to predict measures of attention in neurotypical adults. In dataset three, we show the same predictive network model of autistic traits from dataset one further generalizes to predict measures of social responsiveness in data from the Autism Brain Imaging Data Exchange. In sum, our data suggest that an in-scanner sustained attention challenge can help delineate robust markers of autistic traits and support the continued investigation of the optimal brain states under which to predict phenotypes in psychiatric conditions.

Download



Operating systems in healthcare delivery: body-part and function-based frameworks. We use the term “operating system” to describe the “sequence of decisions and tasks needed to address a health problem” ([1], pg 58). Practically, operating system describes how a patient enters, is sorted by, and arrives at a clinical service line where they receive care for their specific health problem. The sorting function plays a key role in framing the patient’s health problem and in organizing the specialized clinical workforce to address that health problem. Pain medicine operates within a body-part framework (A): when a patient enters the healthcare delivery system, they are asked which body part is painful. Based on the patient’s answer, the healthcare system sorts the patient to the clinical service line that focuses on that body part (e.g., headache clinic versus a spine service line). The painful body part, therefore, is a sorting function that determines not only the clinical service line, but also the skillset of the clinical workforce where the patient is evaluated and treated. Data from the UK Biobank suggests that the body part operating system does not fully capture health problems or pain experience across 34,337 participants. We report biologically grounded pain profiles that could sort connect patients with clinical service lines organized around functional status (B). Each pain profile broadly tracks a different domain of function: pain interference, depression, medical pain, and anxiety. Function-based profiles could see beyond specific body parts and sort patients into a care delivery system that address patients’ overall health. (Adapted from [1], Fig. 5.3. We do not have permission to use these clipart (which we found by google images) but present them as conceptual place holders; we are happy to work with the journal’s graphic designer of choice to improve this figure.)
Large-scale analysis of the UK Biobank reveals 4 distinct pain profiles that track distinct domains of pain experience. Our goal was to define profiles of pain experience at a population level. Across 34,337 participants, we gathered 154 pain-related responses (A) and 100 anatomically defined brain volumes (B), representing a subset of the total UK Biobank (see Supplementary Table 1). We deployed partial least squares canonical correlation as a pattern-learning strategy to trace coherent, biologically driven pain profiles. We identified 4 modes of co-variation that were statistically significant (based on 1000 permutation null distribution shown in grey, panel C). For ease of communication, in subsequent figures we refer to modes 1, 4, 6, and 7 as “pain profiles” 1, 2, 3, and 4, respectively. Each pain profile broadly tracked a distinct domain of pain experience (see Fig. 3)
Four pain profiles capture distinct domains of pain experience. For each significant mode (Modes 1, 4, 6, and 7, hereafter referred to here as pain profile 1, 2, 3, and 4, respectively; see Fig. 2), the left panel plots statistically significant loadings in the context of their null distribution. The center panel provides a sum of statistically significant loadings, grouped by domain of pain experience. The right panel shows the relative brain loadings for each mode, which are further presented in “Results” and in Supplementary Fig. 2. We named each pain archetype based on the symptom domain that contributed most to that mode’s loading: profile 1 broadly tracked pain interference symptomatology, profile 2 tracked depression, profile 3 tracked medical pain, and profile 4 tracked anxiety
Pain profiles capture distinct aspects of population-level covariation. A Diagnosis-wide association studies (Di-WAS, Manhattan plot) relate patterns of pain projections across the population to 1425 clinician-assigned ICD codes across 11 disease classes. B Medication-wide association study (Med-WAS, Manhattan Plot) relate pain projections across the population to 137 medication ATC Level 3 categories across 14 Level 1 ATC Domains. In both Med-WAS and Di-WAS, Pearson’s correlation coefficients (between the pain projection and diagnosis or medication of interest) are shown in units on logarithmic scale of the associated P value. Horizontal lines indicate the significance thresholds at false discovery rate, Bonferroni correction for phenotypes, and Bonferroni correction for phenotypes and asymmetry patterns together labelled FDR, BON, and BON85 (overlapping in each case). For reference, the first pain profile (that broadly tracked pain interference, see Fig. 3) showed unique, strong associations with medications and diagnoses related to painful conditions (opioids and arthropathies). Associations with cardiovascular and metabolic disease were common to all profiles (see Supplementary Figs. 3–5 and Supplementary Table 9). Similarity matrix (C) and a concordance plot (C) show that while each of the four pain profiles broadly tracked distinct symptom domains, the first and second profiles (tracking pain interference and depression) and the third and fourth profiles (tracking medical pain and anxiety) were more correlated across medication, diagnosis, and phenotype-wide association studies. All correlation coefficients shown in C were statistically significant at p < 0.01 level. Overall, D concordance plots across different modes indicated that medication and diagnosis are more similar than phenotype. To facilitate visualization, examples of good, poor, and variable concordance are shown at the base of the figure. E Prescribing behavior across 167,000 participants shows clear trend that patients report taking an antidepressant, opioid, and both (antidepressant and opioid) in relation to depression symptomatology (measured with PHQ-9, severity of depression increases left to right along the x-axis), but not with pain interference (measured with Brief Pain inventory, severity of pain interference increases bottom to top along the y-axis)
Pain can’t be carved at the joints: defining function-based pain profiles and their relevance to chronic disease management in healthcare delivery design

December 2024

·

32 Reads

BMC Medicine

Background Pain is a complex problem that is triaged, diagnosed, treated, and billed based on which body part is painful, almost without exception. While the “body part framework” guides the organization and treatment of individual patients’ pain conditions, it remains unclear how to best conceptualize, study, and treat pain conditions at the population level. Here, we investigate (1) how the body part framework agrees with population-level, biologically derived pain profiles; (2) how do data-derived pain profiles interface with other symptom domains from a whole-body perspective; and (3) whether biologically derived pain profiles capture clinically salient differences in medical history. Methods To understand how pain conditions might be best organized, we applied a carefully designed a multi-variate pattern-learning approach to a subset of the UK Biobank (n = 34,337), the largest publicly available set of real-world pain experience data to define common population-level profiles. We performed a series of post hoc analyses to validate that each pain profile reflects real-world, clinically relevant differences in patient function by probing associations of each profile across 137 medication categories, 1425 clinician-assigned ICD codes, and 757 expert-curated phenotypes. Results We report four unique, biologically based pain profiles that cut across medical specialties: pain interference, depression, medical pain, and anxiety, each representing different facets of functional impairment. Importantly, these profiles do not specifically align with variables believed to be important to the standard pain evaluation, namely painful body part, pain intensity, sex, or BMI. Correlations with individual-level clinical histories reveal that our pain profiles are largely associated with clinical variables and treatments of modifiable, chronic diseases, rather than with specific body parts. Across profiles, notable differences include opioids being associated only with the pain interference profile, while antidepressants linked to the three complimentary profiles. We further provide evidence that our pain profiles offer valuable, additional insights into patients’ wellbeing that are not captured by the body-part framework and make recommendations for how our pain profiles might sculpt the future design of healthcare delivery systems. Conclusion Overall, we provide evidence for a shift in pain medicine delivery systems from the conventional, body-part-based approach to one anchored in the pain experience and holistic profiles of patient function. This transition facilitates a more comprehensive management of chronic diseases, wherein pain treatment is integrated into broader health strategies. By focusing on holistic patient profiles, our approach not only addresses pain symptoms but also supports the management of underlying chronic conditions, thereby enhancing patient outcomes and improving quality of life. This model advocates for a seamless integration of pain management within the continuum of care for chronic diseases, emphasizing the importance of understanding and treating the interdependencies between chronic conditions and pain.


Alterations in Volume and Intrinsic Resting-State Functional Connectivity Detected at Brain MRI in Individuals with Opioid Use Disorder

December 2024

·

13 Reads

·

1 Citation

Radiology

T1-weighted MRI and resting-state functional MRI revealed structural and functional brain alterations in opioid receptor–dense regions, such as the thalamus, right medial temporal lobe, cerebellum, and brainstem, in individuals with opioid use disorder compared with healthy controls.


Overview of study design
Predictive models for infants’ postmenstrual ages were built using structural and functional connectomes separately for term and preterm infants. Brain age gaps (BAGs) were calculated as the difference between predicted and actual postmenstrual age. Structural and functional BAGs were associated with perinatal effects and later cognitive behaviors. Image was created in BioRender. Sun, H. (2024)BioRender.com/m43x607.
Structural and functional connectomes predict postmenstrual age (PMA) in term and preterm infants
a PMA was accurately predicted using structural connectomes for term (blue dots, Pearson’s correlation: r = 0.73, p = 7.39e-74; two-sided) and preterm infants (red dots, Pearson’s correlation: r = 0.67, p = 5.04e-24; two-sided). b PMA was accurately predicted using function connectomes for term (Pearson’s correlation: r = 0.42, p = 3.57e-20; two-sided) and preterm infants (Pearson’s correlation: r = 0.55, p = 6.37e-15; two-sided). Structural (c) and functional (d) connections predicting term and preterm infants’ PMAs. Each heatmap shows the number of edges between each pair of canonical networks during feature selection that were positively (purple) or negatively (green) correlated with postmentural age. VS visual, SM somatomotor, DA dorsal attention, VA ventral attention, LM limbic, FP frontoparietal, DM default mode, SC subcortical. Source data are provided as a Source Data file.
Brain network age prediction
a Structure and (b) Function canonical brain networks age predictions. Within-network connections for multiple networks successfully predicted postmenstrual age in term (blue lines) and preterm (red line) infants. Solid lines indicate significant two-sided Pearson’s correlation at p < 0.05, FDR-corrected, while dashed lines indicate non-significant predictions. Correlations between predicted ages based on within-network connections for term (c) and preterm (d) infants. Heatmaps show the correlation between predicted PMA from within-network connections. The upper triangle shows the correlation between functional ages. The lower triangle shows correlations between structural ages. The diagonal shows the correlations between structural and functional age for a network (p < 0.05; Pearson’s correlation, two-sided; n.s. not significant, box crossed: age not predictable from the within-network connections). VI visual, SM somatomotor, DA dorsal attention, VA ventral attention, LM limbic, FP frontoparietal, DM default mode, SC subcortical. Source data are provided as a Source Data file.
Brain age gaps (BAGs) were associated with maternal effects and toddler behaviors
Maternal mental health, physical health, demographics, and substance use correlate with BAGs for term and preterm infants, controlling for the infant’s postmenstrual age. The BAGs also correlate with several later behaviors in toddlerhood, controlling for the infant’s postmenstrual age. The p-values of two-sided Pearson’s correlation are FDR-corrected. Dashed lines associations were not significant after FDR correction, Solid lines associations were significant after FDR correction. BSID the Bayley Scales of Infant and Toddler Development, CBCL the Child Behavior Checklist, ECBQ the Early Childhood Behavior Questionnaire, Q-CHAT the Quantitative Checklist for Autism in Toddlers. Source data are provided as a Source Data file.
Brain age prediction and deviations from normative trajectories in the neonatal connectome

November 2024

·

42 Reads

Structural and functional connectomes undergo rapid changes during the third trimester and the first month of postnatal life. Despite progress, our understanding of the developmental trajectories of the connectome in the perinatal period remains incomplete. Brain age prediction uses machine learning to estimate the brain’s maturity relative to normative data. The difference between the individual’s predicted and chronological age—or brain age gap (BAG)—represents the deviation from these normative trajectories. Here, we assess brain age prediction and BAGs using structural and functional connectomes for infants in the first month of life. We use resting-state fMRI and DTI data from 611 infants (174 preterm; 437 term) from the Developing Human Connectome Project (dHCP) and connectome-based predictive modeling to predict postmenstrual age (PMA). Structural and functional connectomes accurately predict PMA for term and preterm infants. Predicted ages from each modality are correlated. At the network level, nearly all canonical brain networks—even putatively later developing ones—generate accurate PMA prediction. Additionally, BAGs are associated with perinatal exposures and toddler behavioral outcomes. Overall, our results underscore the importance of normative modeling and deviations from these models during the perinatal period.


BrainEffeX: A Web App for Exploring fMRI Effect Sizes

November 2024

·

29 Reads

Effect size estimation is crucial for power analyses and experimental design, but poses unique challenges in fMRI research due to the complexity of the data and analysis techniques. Here, we utilized large fMRI datasets to obtain precise univariate and multivariate effect size estimates from “typical” fMRI study designs: brain-behavior correlation, task vs. rest, and between-group analyses of functional connectivity and task-based activation maps. We provide an interactive web application for exploring these effect maps (neuroprismlab.shinyapps.io/effect_size_shiny). The app is intentionally designed as a growing resource and we welcome contributions from large (n > 500) datasets.


Figure 3. Hierarchical clustering of the behavioral measures. The labels on the x-axis reflect the test acronyms listed in Table 1. Clustering is based on a Euclidean distance measure for the predictive power for each model across each construct network. Tests that engage the construct networks in the same pattern, indicating that the brain circuits contribute in a similar manner to test score, are clustered together. Behavioral tests are clustered together if their Euclidean distance is less than 0.3. Cluster 1 includes reading (Read), colorword interference (CW), Boston naming test (BNT), and matrix reasoning (MR). Cluster 2 includes verbal learning (VL), finger windows (FW), and verbal fluency 1 (VF1). Cluster 3 includes verbal fluency 2 (VF2) and vocabulary (Vocab). Cluster 4 includes verbal learning delayed recall (VL delay), symbol search (Symbol), coding (Coding), and trail making (Trails).
Figure 4. Predictive power of combined cluster vs. individual behaviors. n=227.
Figure 5. Formation of composite test scores based on individual tests that yield the highest predictive power for individual networks. n=227. The top 25% of behavioral measures (4/16) as indicated by the highest predicted power for each network (from Figure 1) were combined to form composite scores (one score each for the Attention, Perception, Declarative Memory, Language, Cognitive Control, and Working Memory). For each network the top four tests with the highest predictive power were not always the same. The composite score for the attention network was formed from Boston naming test (BNT), reading (Reading), color-word interference (CW), and matrix reasoning (MR) tests. The composite score for the perception network was formed from Boston naming test (BNT), reading (Reading), verbal fluency 1 (VF1), and matrix reasoning (MR). Cluster Dec. Mem. (Declarative Memory) was formed from Boston naming test (BNT), reading (Reading), symbol search (Symbol), and matrix reasoning (MR). Cluster Language was formed from verbal learning delayed recall (VL delay), symbol search (Symbol), vocabulary (Vocab), and matrix reasoning (MR). Cluster Cog. Con. (Cognitive Control) was formed from verbal learning (VL), finger windows (FW), colorword interference (CW), and matrix reasoning (MR). Cluster Work. Mem. (Working Memory)
Testing the Tests: Using Connectome-Based Predictive Models to Reveal the Systems Standardized Tests and Clinical Symptoms Are Reflecting

October 2024

·

32 Reads

Neuroimaging has achieved considerable success in elucidating the neurophysiological underpinnings of various brain functions. Tools such as standardized cognitive tests and symptom inventories have played a crucial role in informing neuroimaging studies, helping to uncover the underlying brain systems associated with these measures. Substantial strides have been taken in developing models, such as connectome-based predictive modeling (CPM), that establish connections between external measures and the human connectome, offering insights into how the functional organization of the brain varies in relation to scores on external measures. Here, we depart from the conventional feed-forward approach and introduce a feed-back approach that allows testing of the tests. Since the inception of cognitive psychology over 60 years ago, cognitive tests have been meticulously developed to measure specific components of cognition. These tests, which have undergone extensive validation and have been standardized and administered to millions, operate on explicit assumptions about the cognitive components they assess. Rather than using external tests to identify the circuits supporting test scores, we a priori define networks of interest and quantify the extent to which these circuits support the test measure. To demonstrate this, we define functional connectivity networks for six cognitive constructs and quantify their contribution to performance across a spectrum of standardized cognitive tests and clinical measures. Employing robust machine learning in a predictive modeling framework, we show how this approach can be used to select tests according to the networks they rely upon. This establishes a biologically grounded metric for test comparison. This approach also yields a brain-driven process for forming composite tests by selecting test combinations that depend on the same proportional brain systems, or for a single network of interest, combining tests with the highest predictive power for that network. This brain-driven approach results in more robust behavioral assessments and enhanced predictive power for the network of interest. We illustrate how this methodology can be applied to evaluate the inclusion of specific sub-tests within a composite score, revealing instances where composite scores are reinforced or weakened by subtest inclusion in terms of the specificity of the brain network they interrogate. The brain-test score modeling approach presented here provides a biologically driven approach to the selection of external cognitive and symptom measures directed at specific brain systems. It opens new avenues of research by providing a framework for the development of new tests and measures guided by quantitative brain metrics.


Altered Functional Connectivity Patterns Associated with Perceived Discrimination in Adolescents

October 2024

·

3 Reads

Perceived discrimination is a pervasive social stressor with significant implications for adolescent mental health and neurodevelopment. Using resting-state functional magnetic resonance imaging (rs-fMRI) data from the Adolescent Brain Cognitive Development (ABCD) Study, we investigated the relationship between perceived discrimination and functional connectivity patterns in the adolescent brain. We applied a Network-Based Statistic (NBS) analysis to 268x268 connectome data from 5,084 participants who provided data on perceived discrimination. Perceived discrimination was coded as a binary variable (Yes/No) based on participants' responses to a validated discrimination scale. Our analysis identified two primary subnetworks associated with perceived discrimination: a positive subnetwork, characterized by increased connectivity between the right prefrontal cortex, left insula, and subcortical regions, and a negative subnetwork, marked by disrupted connectivity between the right prefrontal cortex and left temporal regions. The positive subnetwork suggests enhanced cross-hemispheric communication in adolescents reporting discrimination, while the negative subnetwork indicates reduced cross-hemispheric integration in those reporting no discrimination. Additionally, significant alterations in connectivity across large-scale brain networks were observed. Specifically, adolescents reporting discrimination exhibited increased connectivity between visual processing networks and regions involved in attentional control, while showing decreased connectivity in other network interactions. These findings offer novel insights into the neural mechanisms underlying the effects of perceived discrimination on adolescent brain function, particularly in regions related to emotion regulation, social cognition, and stress responses. Our results underscore the importance of considering discriminatory experiences in the context of adolescent neurodevelopment and mental health.


Citations (39)


... As part of the FIT'NG 2023 meeting, an interdisciplinary panel was convened to discuss the role of developmental neuroimaging in the prediction of neurodevelopmental and psychiatric disorders. This paper is a report-out of a portion of the topics discussed by the panelists, highlighting pressing issues in the use of developmental neuroimaging as a predictive tool (Spann and Scheinost, 2024). ...

Reference:

How Will Developmental Neuroimaging Contribute to the Prediction of Neurodevelopmental or Psychiatric Disorders? Challenges and Opportunities
Applying fetal, infant, and toddler (FIT) neuroimaging to understand mental health
  • Citing Article
  • August 2024

Neuropsychopharmacology: official publication of the American College of Neuropsychopharmacology

... Evaluating a model in an external dataset (i.e., external validation) provides a more robust assessment of replicability and generalizability (Shen et al., 2017;Sui et al., 2020). However, only a minority of neuroimaging studies undertake external validation (Rosenblatt et al., 2024;Yeung et al., 2022). ...

Power and reproducibility in the external validation of brain-phenotype predictions

Nature Human Behaviour

... A recent study, the first to use both fetal and infant scans, shows a significant increase in both intra-and inter-network RSFC from the 30th to the 46th week postmenstrual age. However, this prior study included repeated scans across the birth transition from a small sample (n = 29), and the analysis was restricted to 3 higher-order networks of default mode network, salience network, and the executive control network [35]. ...

Developmental trajectories of the default mode, frontoparietal, and salience networks from the third trimester through the newborn period

... A recent systematic review and meta-analysis of seven studies involving 226,541 participants estimated that the risk of urolithiasis in MAFLD patients is 1.73 times higher than in healthy controls [10]. Although many studies have attempted to reveal the relationship between MAFLD and urolithiasis, it is difficult to infer the true causal relationship due to some confounding factors [10][11][12][13][14]. Mendelian randomization (MR) is a genetic approach that leverages data from large-scale genome-wide association studies (GWAS) to estimate disease risk [15]. MR studies employ genetic variants as instrumental variables to derive valid causal inferences, effectively minimizing the influence of confounding factors [16]. ...

The brain structure, inflammatory, and genetic mechanisms mediate the association between physical frailty and depression

... This is in contrast to the KRR model implemented during the meta-matching stacking process, which was trained using the DNN-generated cognitive, health, and behavioral as inputs. KRR was used as a baseline model as it is widely used and has repeatedly been shown to work well for functional connectivitybased behavioral and demographic prediction (24,29,38,41,(81)(82)(83). In a recent work, we have further demonstrated that meta-matching with stacking is superior to a classic transfer learning approach (84). ...

Connectome caricatures: removing large-amplitude co-activation patterns in resting-state fMRI emphasizes individual differences

... Compared to traditional methods such as the sliding-window approach or quasi-periodic pattern analysis with a predefined window size (Majeed et al., 2011), CPCA demonstrates superior precision in detecting the individual dynamic patterns without the need for a fixed arbitrary windowing across individuals from different developmental stages. This makes CPCA particularly flexible and suitable for use with neurodevelopmental studies (Foster & Scheinost, 2024). It's important to note when studying dynamics, particularly those at the individual level, data quantity remains a critical factor. ...

Brain states as wave-like motifs
  • Citing Article
  • April 2024

Trends in Cognitive Sciences

... Benjamin Libet and his colleagues simply noted that there is not yet a complete causal explanation for the movement-and the missing variables in the equation can be conscious or unconscious processes (see also Dominik et al., 2023;Nestor, 2019). We have already seen in this chapter why the unconscious/conscious distinction is so important for law. ...

The tip of the iceberg: A call to embrace anti-localizationism in human neuroscience research

... PCA was applied on every individual wavelength to extract specific wavelengths that were characteristic in foliar symptomatic appearances. PCA was applied to the training set only to avoid overfitting due to data leakage [33]. Then, an incremental feature selection method was conducted to explore the possible combinations of the PCA-determined key wavelengths, and the optimal wavelength combinations were determined on each of the seven machine learning models based on the F1-score metric. ...

Data leakage inflates prediction performance in connectome-based machine learning models

... Some instances of dataset shift are unavoidable, but ongoing efforts to improve MR sequence and measurement harmonization, such as the Common Measures for Mental Health Science 59 , may reduce dataset shift. Still, unharmonized, or even negatively correlated, measures can be used in external validation 60,61 , such as predicting a clinical measure of attention-deficit/hyperactivity disorder from a sustained attention network despite a negative correlation between the clinical and sustained attention measures 60 . Finally, dataset shift can be mitigated with large, representative samples 57,62 . ...

Brain-phenotype predictions can survive across diverse real-world data

... The problem of missing data in large brain datasets is a new topic receiving increasing attention. In the past year, novel methods, such as predictive mean matching [15] and a graph neural network [25], have been proposed for imputing missing functional connectomes for individual subjects [15] and missing morphometry values from an entire dataset [25]. These methods have improved performance on downstream tasks. ...

Rescuing missing data in connectome-based predictive modeling