Research
Machine Learning Methods, Computational Biology, and Digital Medicine

Our research spans three connected directions: developing machine learning methods, understanding biological systems, and translating molecular, imaging, and clinical data into digital medicine.

Machine Learning Methods
Self-supervised Learning
Developing methods that learn useful representations from complex data with limited labels, with applications to omics and imaging.
Multimodal Integration
Developing learning methods that integrate heterogeneous data while preserving shared and modality-specific information.
Continuous Space and Time Modeling
Modeling biological variation across continuous spatial coordinates and time to reconstruct tissue organization and dynamic processes.
Computational Biology
Multiscale Biological Modeling
Reconstructing molecular, cellular, and tissue organization from noisy or incomplete measurements to understand biological systems and disease.
Target Discovery
Connecting spatial omics, cellular states, and perturbation data to investigate disease mechanisms and identify molecular targets.
Digital Medicine
Disease Phenotype Modeling
Integrating pathology, molecular profiles, and clinical phenotypes to characterize disease variation and investigate biomarkers for patient stratification.
Drug Discovery
Learning from genetic, perturbation, and imaging data to predict compound function and support drug screening and repurposing.