We develop multimodal deep learning and representation learning methods to integrate electronic health records, multi-omics, and digital biomarkers for disease risk stratification, outcome prediction, and clinical decision support. Our work emphasizes data-efficient, interpretable, and robust models that can be reliably deployed in real-world healthcare settings.
We build scalable machine learning and statistical methods to identify molecular targets and mechanisms from high-dimensional biological data. Our goal is to translate molecular insights into actionable hypotheses for disease understanding, drug discovery and precision medicine at the population-level.
We investigate the determinants of brain resilience, the ability to maintain cognitive function despite aging and pathology—by integrating molecular, clinical, and epidemiological data. We aim to identify neuroprotective signatures in neurodegenerative diseases and uncover targets for therapeutic intervention.