Current Research and Scholarly Interests
My research program advances precision medicine by integrating artificial intelligence, real-world healthcare data, wearable biosensing, multi-omics, and medical imaging. A central goal is to convert complex clinical, physiological, and molecular information into actionable models that can predict disease risk, identify mechanistic subtypes, guide intervention, and support scalable implementation across health systems.
In population health and clinical AI, my work leverages large-scale electronic health record and health information exchange datasets to develop predictive models for disease onset, progression, clinical events, healthcare utilization, and mortality. These models support earlier risk recognition, more efficient care navigation, and personalized intervention strategies. Increasingly, this work is moving toward multimodal foundation models and digital twin frameworks that can represent longitudinal patient histories, integrate new data over time, and support dynamic clinical decision-making.
A major and expanding focus of my research is wearable AI. We develop approaches to use biosensor data, including heart rate, heart rate variability, activity, sleep, photoplethysmography-derived signals, and other continuous physiological measurements, to monitor health trajectories in real time. By integrating wearable signals with electronic health records, molecular profiles, and clinical outcomes, we aim to detect early deviation from baseline, identify impending clinical deterioration, and enable personalized digital twin analysis. This work is relevant for pregnancy disorders, pediatric disease monitoring, cardiometabolic risk surveillance, and population-scale preventive health.
In molecular diagnostics and systems biology, my laboratory develops LC-MS/MS-based proteomic and metabolomic platforms for liquid biopsy and tissue-based profiling. These approaches enable quantitative measurement of thousands of proteins, metabolites, and lipid mediators from blood, tissue, and other clinically relevant biospecimens. We apply these technologies to discover and validate biomarkers for pregnancy complications, preterm birth, preeclampsia, abnormal uterine bleeding, pediatric inflammatory diseases such as Kawasaki disease, and other disorders with significant unmet diagnostic needs.
A major area of current emphasis is maternal-fetal and pediatric precision medicine. Our work integrates circulating protein biomarkers, cfDNA methylation, metabolomics, wearable physiological signals, and clinical phenotypes to define disease trajectories, identify high-risk patients before clinical deterioration, and uncover mechanisms underlying pregnancy and pediatric disorders. This includes efforts to develop early blood-based tests for preterm birth and preeclampsia, AI tools for Kawasaki disease diagnosis and treatment response prediction, and multimodal clocks of reproductive, placental, and pediatric health.
In medical imaging and computer-aided analytics, we develop deep-learning approaches to extract clinically meaningful signals from echocardiograms, whole-slide pathology images, and spatial molecular data. These imaging models are combined with wearable, molecular, and EHR-derived features to create integrated risk models that move beyond single-modality prediction toward comprehensive biological and clinical interpretation.
Across these domains, the unifying goal is translational implementation: building AI, wearable, multi-omic, and imaging technologies that can move from discovery to validation, from prediction to intervention, and from specialized research settings to real-world clinical and population health applications. My long-term vision is to create integrated learning health systems in which multimodal data, continuous physiological monitoring, biological mechanisms, and AI-driven models inform more precise, preventive, and equitable healthcare.
