Research
I translate advanced neuroimaging and cognitive neuroscience into scalable, real-world tools for mental health and decision support.
Wearable fNIRS for precision mental health
A low-cost, wireless functional near-infrared spectroscopy (fNIRS) platform for objective assessment and repeated monitoring of brain function in naturalistic settings. The system pairs portable hardware with cloud-based signal processing, behavioral tasks, and ecological momentary assessment (EMA). This makes dense-sampling, individual-level neuroimaging practical outside the lab.
Neuromonitoring-guided intervention in ADHD
Characterizing neurobiological heterogeneity in ADHD with multimodal wearable neuroimaging and behavior, and using neural markers to personalize intervention. This work combines task-based and naturalistic fNIRS paradigms with neurofeedback-informed working-memory training. The aim is objective markers of executive function and treatment response, supporting precision psychiatry beyond symptom-based diagnosis.
Foundation models for optical neuroimaging
Self-supervised representation learning for fNIRS time series, enabling brain-state discovery that generalizes across tasks, populations, and contexts. The work spans large-scale aggregation of wearable fNIRS data, masked and contrastive learning objectives, and a phenotype-first modeling strategy that keeps the learned representations interpretable and clinically relevant.
Naturalistic, at-home neuroimaging
Measuring the brain where people actually live. In a pilot of self-initiated, ambulatory prefrontal fNIRS recordings during everyday activities paired with ecological momentary assessment (98 sessions over 24 days), we examine task- and mood-dependent prefrontal functional connectivity, alongside dense-sampling deep phenotyping in ADHD.
Clinical translation: Alzheimer’s disease and presurgical mapping
As lead analyst with the Stanford Alzheimer’s Disease Research Center, I study wearable fNIRS during NIH Toolbox cognitive tasks in an amyloid-characterized cohort, linked to amyloid and tau PET, plasma biomarkers, and cognition. A separate pilot compares whole-head and wearable fNIRS for covert language mapping, toward a bedside alternative to Wada testing and fMRI for presurgical language lateralization.
Timing, action, and the motor system
How context shapes the neural control of timed action. Using fNIRS, EEG, and deep learning, this work examines how auditory cues, task structure, and study design modulate activity in primary and supplementary motor cortex during rhythmic behavior, including in cochlear-implant users.
Neurodiversity and decision-making under risk
How neurodiverse individuals process risk, loss aversion, and feedback frequency in financial and everyday decisions. The project brings wearable neuroimaging into decision tasks grounded in behavioral economics, bridging neuroscience and real-world decision support.
Clinical neuroimaging in vulnerable populations
Objective neurophysiological markers of stress and emotion regulation in clinically complex and underserved groups, including PTSD, adverse childhood experiences (ACE), and Prader–Willi syndrome. Current work links prefrontal responses during acute stress to childhood-adversity load and heart-rate variability. The emphasis is on ethical, scalable neuroimaging that can inform intervention and monitoring.
Methods and open science in fNIRS
Contributing to community efforts that improve rigor and reproducibility in fNIRS research, including the multi-lab FRESH reproducibility initiative the consensus-based fNIRS Glossary Project, and a multiverse (specification-curve) replication of fNIRS findings on non-adjacent dependency learning.