We develop statistical models to:
identify associations between complex-structured data (imaging, spatial-genomic, geospatial, digital) from multi-platform data sources,
predict clinical characteristics of interest by integrating biomarkers generated from complex-structured data, and
understand the biological relevance and implications of our findings.
Our methods are developed in the context of various diseases such as cancer and neurodegenerative (Alzheimer’s) diseases.
We develop innovative statistical methodology that leverages structural and biological information in the data. To address the challenges of handling complex-structured data, we develop and employ methods at the interface of hierarchical Bayesian modeling, variable selection, geometric/functional data analysis and spatial statistics.