Applications to Genetics and Health Data

Genetic and clinical studies must often combine heterogeneous data while accommodating imperfect instruments, population differences, and limits on patient-level data sharing. My work brings robust causal inference, federated learning, and transfer learning to these problems, with the aim of producing biomarker findings and risk predictions that remain reproducible across biobanks, hospitals, and patient populations.

Authorship notation Name supervised student # equal contribution * alphabetical ordering corresponding author