Research

My research program develops statistical and computational methods that integrate large-scale, multi-source datasets to address problems in public health and medicine, with an emphasis on health equity, statistical genetics, and scalable, open-source software. The work spans methods development, large-scale data analysis, and the release of tools that the broader research community can use directly.

Health equity in disease risk prediction

A central theme of my work is making polygenic risk prediction accurate and equitable across diverse ancestry groups, where methods trained primarily in European-ancestry cohorts often underperform. I develop Bayesian and ensemble-learning methods that borrow information across populations to improve ancestry-specific polygenic risk scores (PRS), and I build platforms that make these methods usable at scale.

• MUSSEL — Enhanced Bayesian polygenic risk prediction leveraging information across multiple ancestry groups. Cell Genomics, 2024. [Paper] [Code]
• PennPRS — A centralized cloud-computing platform for efficient PRS model training in precision medicine. medRxiv, 2025. [Preprint] [Platform]

Causal inference and Mendelian randomization

I develop Mendelian randomization (MR) methods that recover the effects of latent or hard-to-measure exposures by integrating GWAS summary statistics across multiple observed biomarkers, improving power and robustness over single-biomarker approaches.

• MRLE — Mendelian randomization analysis using multiple biomarkers of an underlying common exposure. Biostatistics, 2024. [Paper] [Code]
• CaLMR — Bayesian MR analysis for latent exposures leveraging GWAS summary statistics for co-regulated traits. medRxiv, 2024. [Preprint] [Code]

High-dimensional and graph-informed inference

I build methods for high-dimensional, complex-structured data that incorporate known biological structure — such as gene networks — to improve the detection of disease-associated signals.

• T2-DAG — A powerful test for differentially expressed gene pathways via graph-informed structural equation modeling. Bioinformatics, 2022. [Paper] [Code]

Risk prediction for population and public health

I apply statistical modeling to translate large-scale data into individual- and community-level risk assessments that inform public-health decision making.

• Individual and community-level risk for COVID-19 mortality in the United States. Nature Medicine, 2021. [Paper]

Funding

My research is supported by an NIH Pathway to Independence Award (K99/R00) from the National Human Genome Research Institute (NHGRI), for which I serve as Principal Investigator, developing methods for multi-ancestry disease risk prediction that integrate genetic and non-genetic risk factors across data sources. [NIH RePORTER]

For a full list of publications, see the Publications page; software and platforms are available on the Software page.