I am mainly interested in developing novel analytical tools to address statistical challenges (including high-dimensionality, paucity of accurate labels, heterogeneity, privacy constraints, etc) in analyzing large-scale electronic health record (EHR) and biobank data. Relevant statistical research areas include: high-dimensional statistics, semiparametric inference, federated learning, data-fusing, semi-supervised learning, transfer learning, distributionally robust optimization, model-X inference, and selective inference. My developed methods have been applied for EHR phenotyping and model evaluation, multi-source data aggregation, genetic risk modeling, phenotype-genotype association studies, clinical knowledge extraction, etc.