About

Kaustubh Supekar is a Clinical Assistant Professor in the Department of Psychiatry and Behavioral Sciences at Stanford University. He is affiliated with the Center for Artificial Intelligence in Medicine & Imaging (AIMI) at Stanford. His work focuses on the application of artificial intelligence to medicine and imaging, contributing to research in healthcare and medical imaging through his role at AIMI. The page highlights his position within Stanford's academic community and his involvement in advancing AI-driven healthcare solutions.

Research topics

• Neuroscience
• Psychiatry
• Psychology
• Cognitive psychology
• Clinical psychology
• Genetics
• Biology

Selected publications

• Artificial Intelligence in Obsessive-Compulsive Disorder: A Systematic Review

  Current Treatment Options in Psychiatry · 2025-06-14 · 6 citations

  reviewOpen access

  Purpose of Review: Obsessive-compulsive disorder (OCD) is a chronic and disabling condition, often leading to significant functional impairments. Despite its early onset, there is an average delay of 17 years from symptom onset to diagnosis and treatment, resulting in poorer outcomes. This systematic review aims to synthesize current findings on the application of AI in OCD, highlighting opportunities for early symptom detection, scalable therapy training, clinical decision support, novel therapeutics, computer vision-based approaches, and multimodal biomarker discovery. Recent Findings: While previous reviews focused on biomarker-based OCD detection and treatment using machine learning (ML), the findings of the current review add information about novel applications of deep learning technology, specifically generative artificial intelligence (GenAI) and natural language processing (NLP). Among the included 13 articles, most studies (84.6%) utilized secondary data analyses, primarily through GenAI/NLP. Nearly 77% of these studies were published in the past two years, with high quality of evidence. The primary focus areas were enhancing treatment and management, and timely OCD detection (both 38.5%); followed by AI tool development for broader mental health applications. Summary: AI technologies offer transformative potential for improvements related to OCD if diagnosis occurs earlier after onset; thereby lessening the consequential economic burden. Prioritizing investment in ethically sound AI research could significantly improve OCD outcomes in mental health care. Supplementary Information: The online version contains supplementary material available at 10.1007/s40501-025-00359-8.

  PublisherOA PDFDOI

• Glutamate, Contextual Insensitivity, and Disorganized Speech in First-Episode Schizophrenia: A 7T Magnetic Resonance Spectroscopy Study

  Biological Psychiatry Global Open Science · 2025-08-13 · 2 citations

  articleOpen access

  In people with schizophrenia, formal thought disorder is a core symptom that emerges early and persists into chronic stages despite treatments. It manifests as disorganized speech and is often associated with poor long-term outcomes. A key feature of this disorganization is an impairment in the buildup and use of context provided by preceding words when choosing upcoming words. Recent work has shown that spoken words are less predictable based on global linguistic context in schizophrenia, but the neural basis of this remains unknown. Glutamate dysfunction in the anterior cingulate cortex has long been implicated in schizophrenia, but its connection to behavioral impairments remains unclear. In this study, we investigated the relationship between linguistic contextual sensitivity and glutamate level in the dorsal anterior cingulate cortex in 39 patients with first-episode psychosis (33 men) and 33 sociodemographically matched healthy control participants (22 men). Contextual sensitivity was measured using a large language model (GPT-3), and glutamate levels were measured using 7T magnetic resonance spectroscopy. We found a significant interaction between diagnosis and glutamate level in predicting contextual sensitivity: Patients with lower glutamate levels had poor contextual sensitivity, a relationship not seen in healthy control participants. Glutamate variation was specifically explained by contextual sensitivity after controlling for other clinical and language variables, underscoring the robustness and specificity of this association. These results highlight a potential glutamatergic basis for disorganized speech in schizophrenia and suggest that contextual sensitivity in speech could reflect anterior cingulate glutamate variations in early psychosis. Disorganized speech is a common and disabling symptom of schizophrenia, but its biological roots remain unclear. In this study, we found that in untreated first-episode psychosis, lower levels of glutamate—the brain’s major excitatory neurotransmitter—in the anterior cingulate cortex were linked to greater difficulty processing context in speech. These findings suggest that speech disruptions in schizophrenia may stem from specific brain chemistry changes early in the illness.

  PublisherDOI

• Shared and distinct neural signatures of inhibitory control deficits in attention-deficit/hyperactivity disorder and autism

  bioRxiv (Cold Spring Harbor Laboratory) · 2025-10-12

  preprintOpen accessSenior author

  Abstract Background Attention-deficit/hyperactivity disorder (ADHD) and autism spectrum disorder (ASD) share deficits in inhibitory control, yet direct comparisons of their neurobiological underpinnings remain sparse and are often confounded by comorbidity. We examined shared and distinct abnormalities in brain activation and circuit connectivity associated with inhibitory control in non-comorbid ADHD and ASD. Methods Brain activity and hyperdirect pathway connectivity during inhibitory control were examined in children with non-comorbid ADHD (age: 11.1±0.24y) and children with non-comorbid ASD (age: 11.1±0.39y), compared with typically developing (TD) children (age: 10.9±0.09y), and benchmarked against healthy adults (age: 24.1±1.26y), using fMRI stop-signal task data. Results Both ADHD and ASD groups showed reduced activation in salience and frontoparietal network regions alongside increased activation in default mode network (DMN) regions, consistent with triple-network dysfunction. These activation deficits were more pronounced in ASD, whose activation profile also diverged most from adults. At the circuit level, only ASD showed abnormal hyperdirect connectivity, involving insula-STN and preSMA-STN pathways and further reflected in a composite insula/IFG-STN connectivity measure, indicating broader cortical-STN disruption. Importantly, across all children, reduced salience/frontoparietal relative to DMN engagement correlated with greater parent-reported inhibitory control deficits, and in ASD, individual differences in insula/IFG-STN connectivity showed the same association, highlighting the behavioral relevance of both network- and circuit-level abnormalities. Conclusions ADHD and ASD share network-level but differ in circuit-level abnormalities underlying inhibitory control deficits. Together, these insights advance the neurobiology of inhibitory control in ADHD and ASD and provide potential targets for neuroscience-informed interventions addressing both shared and disorder-specific mechanisms.

  PublisherOA PDFDOI

• Reply to Lockhart et al.: Advancing the understanding of sex differences in functional brain organization with innovative AI tools

  Proceedings of the National Academy of Sciences · 2025-01-02

  articleOpen access
  PublisherOA PDFDOI

• Robust and replicable functional brain signatures of 22q11.2 deletion syndrome and associated psychosis: a deep neural network-based multi-cohort study

  Molecular Psychiatry · 2024-04-12 · 4 citations

  article1st author
  PublisherDOI

• Deep learning models reveal replicable, generalizable, and behaviorally relevant sex differences in human functional brain organization

  Proceedings of the National Academy of Sciences · 2024-02-20 · 55 citations

  articleOpen access

  Sex plays a crucial role in human brain development, aging, and the manifestation of psychiatric and neurological disorders. However, our understanding of sex differences in human functional brain organization and their behavioral consequences has been hindered by inconsistent findings and a lack of replication. Here, we address these challenges using a spatiotemporal deep neural network (stDNN) model to uncover latent functional brain dynamics that distinguish male and female brains. Our stDNN model accurately differentiated male and female brains, demonstrating consistently high cross-validation accuracy (>90%), replicability, and generalizability across multisession data from the same individuals and three independent cohorts (N ~ 1,500 young adults aged 20 to 35). Explainable AI (XAI) analysis revealed that brain features associated with the default mode network, striatum, and limbic network consistently exhibited significant sex differences (effect sizes > 1.5) across sessions and independent cohorts. Furthermore, XAI-derived brain features accurately predicted sex-specific cognitive profiles, a finding that was also independently replicated. Our results demonstrate that sex differences in functional brain dynamics are not only highly replicable and generalizable but also behaviorally relevant, challenging the notion of a continuum in male-female brain organization. Our findings underscore the crucial role of sex as a biological determinant in human brain organization, have significant implications for developing personalized sex-specific biomarkers in psychiatric and neurological disorders, and provide innovative AI-based computational tools for future research.

  PublisherOA PDFDOI

• Novel Neural Network Models for Predicting Mental Health Outcomes in the U.S. Youth Population

  Journal of Student Research · 2024-02-29

  articleOpen accessSenior author

  Anxiety and depression are two of the most pressing mental health issues, particularly among young adults. Given the success of neural networks for predictive modeling, we developed novel neural network models for classifying anxiety and depression using Substance Abuse and Mental Health Services Administration’s Mental Health Client- Level Data (SAMHSA MH-CLD) on 382,174 young adults (15 to 24 years old). The SAMHSA MH-CLD included mental health and general background data collected in 2020 for individuals reporting to state-accredited hospital service centers across the United States. The neural networks were trained on 10% randomized k-folds of the dataset and tested on the remaining 90% for each fold. We found that neural network models predicted anxiety and depression with high accuracy (91.5% to 94.2% accuracy, 8.4% to 3.1% loss), outperforming conventional statistical models. Additionally, for all tested variable sets, our neural network model outperformed expectations for the average therapist consultation (46% to 50% accuracy), as reported previously. For the optimal neural network model with the highest accuracy, the variables most correlated with anxiety and depression were age, education, gender, race, employment, marital status, and stressor events, not accounting for redundant and minimally correlated variables. The effectiveness of our neural network model indicates that it can be implemented alongside therapists in clinical environments to improve psychiatric diagnosis among young adults.

  PublisherOA PDFDOI

• Neuroanatomical, transcriptomic, and molecular correlates of math ability and their prognostic value for predicting learning outcomes

  Science Advances · 2024-05-31 · 2 citations

  articleOpen accessCorresponding

  Foundational mathematical abilities, acquired in early childhood, are essential for success in our technology-driven society. Yet, the neurobiological mechanisms underlying individual differences in children's mathematical abilities and learning outcomes remain largely unexplored. Leveraging one of the largest multicohort datasets from children at a pivotal stage of knowledge acquisition, we first establish a replicable mathematical ability-related imaging phenotype (MAIP). We then show that brain gene expression profiles enriched for candidate math ability-related genes, neuronal signaling, synaptic transmission, and voltage-gated potassium channel activity contributed to the MAIP. Furthermore, the similarity between MAIP gene expression signatures and brain structure, acquired before intervention, predicted learning outcomes in two independent math tutoring cohorts. These findings advance our knowledge of the interplay between neuroanatomical, transcriptomic, and molecular mechanisms underlying mathematical ability and reveal predictive biomarkers of learning. Our findings have implications for the development of personalized education and interventions.

  PublisherOA PDFDOI

• 23. The Effects of Methylphenidate on Spontaneous Fluctuations in Reward and Cognitive Control Networks in Children With Attention-Deficit/Hyperactivity Disorder -Randomized Controlled Studies in Two Independent Cohorts

  Biological Psychiatry · 2023-04-10

  articleOpen access
  PublisherOA PDFDOI

• Methylphenidate Enhances Spontaneous Fluctuations in Reward and Cognitive Control Networks in Children With Attention-Deficit/Hyperactivity Disorder

  Biological Psychiatry Cognitive Neuroscience and Neuroimaging · 2022-10-23 · 18 citations

  articleOpen access

  BACKGROUND: Methylphenidate, a first-line treatment for attention-deficit/hyperactivity disorder (ADHD), is thought to influence dopaminergic neurotransmission in the nucleus accumbens (NAc) and its associated brain circuitry, but this hypothesis has yet to be systematically tested. METHODS: We conducted a randomized, placebo-controlled, double-blind crossover trial including 27 children with ADHD. Children with ADHD were scanned twice with resting-state functional magnetic resonance imaging under methylphenidate and placebo conditions, along with assessment of sustained attention. We examined spontaneous neural activity in the NAc and the salience, frontoparietal, and default mode networks and their links to behavioral changes. Replicability of methylphenidate effects on spontaneous neural activity was examined in a second independent cohort. RESULTS: Methylphenidate increased spontaneous neural activity in the NAc and the salience and default mode networks. Methylphenidate-induced changes in spontaneous activity patterns in the default mode network were associated with improvements in intraindividual response variability during a sustained attention task. Critically, despite differences in clinical trial protocols and data acquisition parameters, the NAc and the salience and default mode networks showed replicable patterns of methylphenidate-induced changes in spontaneous activity across two independent cohorts. CONCLUSIONS: We provide reproducible evidence demonstrating that methylphenidate enhances spontaneous neural activity in NAc and cognitive control networks in children with ADHD, resulting in more stable sustained attention. Our findings identified a novel neural mechanism underlying methylphenidate treatment in ADHD to inform the development of clinically useful biomarkers for evaluating treatment outcomes.

  PublisherDOI

Frequent coauthors

• Vinod Menon

  Neurosciences Institute

  103 shared

• Srikanth Ryali

  24 shared

• Weidong Cai

  18 shared

• Tianwen Chen

  Hangzhou First People's Hospital

  15 shared

• Yoshifumi Mizuno

  University of Fukui

  14 shared

• Lucina Q. Uddin

  11 shared

• Akemi Tomoda

  University of Fukui

  10 shared

• Lena Palaniyappan

  Western University

  10 shared

Education

• Ph.D.

  Stanford University