Data integration

The development of new statistical methods for integrative analyses of multiple data sources empowers us to answer meaningful scientific questions that could not have been addressed with a single data source. I am particularly interested in problems related to heterogeneity of data sources, dependence between data sources, and computational efficiency and data privacy of distributed methods.
   - applications in neuroimaging and metabolomics

Metabolomics

Metabolites are small molecules in biological systems that are end-products of cell metabolism. The study of the metabolome, the collection of metabolites in a tissue, can elucidate the biological mechanism of many metabolism-related diseases, such as diabetes and obesity. The complex networked structure of metabolites and the high proportion of missingness present challenges and opportunities for improved statistical analysis.
   - collaboration with Jian Kang and Michael Boehnke at the University of Michigan
​   - collaboration with the Data Management and Modeling Core of the Children's Environmental Health and Disease Prevention Center of the University of Michigan
   - collaboration with David Muddiman and his lab at North Carolina State University

Wearable devices

Wearable devices, such as the Apple Watch and Fitbit, are ubiquitous in daily life and have empowered people to take charge of their health in unprecedented ways. They repeatedly collect health-related data such as movement and heart rate at high speed, and combined with questionnaires and diaries are a powerful tool for medical discovery. I have recently become particularly interested in the characterization of sleep and its quality, and in studying the determinants of high and low quality sleep.
   - collaboration with Erica Jansen and the ELEMENT group at the University of Michigan
   - collaboration with Lan Luo at the University of Iowa
​