Area of Research

Brain–Behavior Triangulation Under Real-World Heterogeneity

My research asks how brain–behavior relationships can be characterized in heterogeneous real-world settings, where cognition, symptoms, brain systems, biomarkers, and clinical context are measured across different scales and sources. I use a triangulation-based approach: rather than relying on a single task, cohort, modality, or clinical measure, I combine complementary data sources to test whether candidate brain–behavior patterns are robust across contexts, context-specific, or individualized. I call this emerging direction computational neuroecology—using computational models and multi-source data to study how brain–behavior relationships manifest in the settings where cognition and clinical outcomes actually emerge. Aging, late-life neuropsychiatric symptoms, and Alzheimer's disease and related dementias (ADRD) are primary application areas, but the broader goal is to build interpretable models that connect cognitive neuroscience, multimodal neuroimaging, and clinical data science.

Multi-modal Neuroimaging & Data Fusion

No single modality captures the full complexity of brain disorders and aging. I develop and apply computational frameworks for integrating structural MRI, PET, fluid biomarkers, electronic health records, and cognitive assessments. I have extended and applied SuperBigFLICA, a multi-cohort framework inspired by linked independent component analysis, to fuse multimodal neuroimaging data across large, heterogeneous datasets. My focus is on identifying multi-modal patterns that replicate across cohorts and inform clinical outcomes and individual-level trajectories.

Clinical Data Science, EHR & Biomarkers

Complementing neuroimaging, I use fluid biomarkers and apply NLP—including LLM-based approaches—to heterogeneous text sources including EHR (ICD codes and clinical notes), free-text self-reports, and interview transcripts to extract clinical and behavioral phenotypes. At MGH, I contributed to large-scale neuroimaging analysis tools (including SynthSeg+, published in PNAS) and applied NLP methods to clinical text in dementia research. At McLean, I integrate these data streams with neuroimaging to study neuropsychiatric symptoms, disease staging, and cognitive trajectories in aging and ADRD.

Spatial Navigation & Cognitive Neuroscience

My doctoral research established that travel direction is a fundamental and independent component of human navigation—distinct from the well-studied head direction signal—using fMRI and novel psychophysical paradigms. Spatial navigation is among the earliest cognitive domains affected in Alzheimer's disease, making the behavioral and neural measures from this work potentially relevant to early detection and intervention.

Individual Differences Across Populations

A thread running through all my research is attention to individual differences—across sex, culture, age, and clinical status. During my doctoral work, I examined how handedness and biological sex influence spatial cognition using both human participants and computational models, and conducted large-scale analyses of navigation behavior across countries using data from nearly 770,000 participants.