Computational Modeling of Multimodal Data in Aging, Somatic Mosaicism, and Cardiovascular Disease
Our research uses computational modeling to integrate multimodal biomedical data—including human multi-omics, medical imaging, and EHR data—to uncover disease mechanisms linking aging and somatic mosaicism to cardiovascular and other age-related diseases, improve clinical phenotyping and risk stratification, and identify and prioritize therapeutic targets.
Key areas of interest:
Aging, somatic mosaicism, and disease: We investigate somatic mutations in blood—including clonal hematopoiesis—and other tissues, and their relationships with aging and disease. We integrate human genetics, multi-omics, and experimental evidence to distinguish causal mechanisms from associations and prioritize targets with potential for therapeutic intervention.
Multimodal phenotyping and machine learning: We use deep learning to analyze human and murine imaging data and LLMs to derive phenotypes from EHR data. We integrate these modalities with genomics and other omics data to uncover biological mechanisms and improve disease characterization.
Omics and statistical methods: We develop computational and statistical methods for multi-omics integration, cross-platform harmonization, quality assessment, and reproducible biological discovery.
Our research is enabled by access to extensive human and murine datasets. We are deeply grateful to our collaborators and cohort investigators for their partnership and trust.
Congratulations to Qiayi, Linke, Antony, and Kathy on their conference presentations! Qiayi's work, The virtual longevity clinic: a multi-agent LLM estimates biological age, was selected for an oral presentation at the 2026 Biomarkers of Aging Conference; Linke's work, Machine learning cross-platform proteomic imputation enables protein quality scoring and replication of epidemiological associations, was selected for an oral presentation at the 2026 CHARGE Meeting, where she also received a travel award; and Antony's work, From proteins to pathways: a biologically informed neural network for risk factors and incident coronary artery disease, and Kathy's work, Shared genetic architecture across cardiovascular-kidney-metabolic syndrome, were both selected for moderated digital poster presentations at the upcoming American Heart Association (AHA) Scientific Sessions!
Multiple papers accepted by Cell, European Heart Journal, Circulation, JACC: CardioOncology, Nature Communications (x2), and Hypertension!
Welcome Rong and Rusty to the lab!
Congratulations to Meghana, Anthea, and Haodong on their paper published in Science Translational Medicine! The study was also featured by Harvard Medical School and Mass General Brigham, with additional coverage from Mass General Brigham on LinkedIn. [paper]
Welcome Aram to the lab!