Data Science and AI

Data science and artificial intelligence are revolutionizing diabetes research by enabling analysis of large, complex datasets to uncover new insights into disease progression and response to treatments. Our research program leverages biostatistics, bioinformatics, machine learning algorithms, and real-world evidence to accelerate the development of personalized treatment strategies and to improve outcomes for the millions of people living with diabetes worldwide.

Biostatistics 
The biostatistics focus within the AI and Data Science program provides rigorous methodologies for scientific inquiry, emphasizing development and application of statistical methods tailored to the challenges of diabetes research. Our biostatisticians specialize in designing robust clinical trials, development of innovative approaches for analysis of longitudinal data, and development of predictive models to better understand the complex, multifactorial nature of diabetes and associated complications. The team works collaboratively with clinicians and data scientists to ensure that models are statistically sound and clinically interpretable. Other areas of expertise include analysis of electronic health records, continuous glucose monitor and other device data, basic science experiments, and patient-reported outcomes.

Multi-omics Integration 
Members of this program use cutting-edge approaches to understand diabetes through analysis and integration of diverse layers of molecular data, including genomics, epigenetics, transcriptomics, proteomics, metabolomics, radiomics, as well as spatial omics and exposomics. We develop algorithms and pipelines to harmonize heterogenous datasets, addressing the unique challenges of scale, dimensionality, sparseness, and heterogeneity that arise when combining molecular profiles with clinical phenotypes and longitudinal outcomes. Our integration frameworks map the complex molecular interactions underlying diabetes pathophysiology, insulin resistance, and complications, while accounting for genetic variations and environmental factors. This program bridges the gap between molecular biology and clinical translation by developing models that can stratify patients into phenotypic subgroups, predict treatment response, and guide precision medicine approaches.

AI 
We harness the power of advanced machine learning methodologies to transform diabetes care by developing state-of-the-art algorithms to analyze and combine imaging data, omics data, and clinical data. This research emphasis includes development of computational frameworks to analyze multidimensional datasets while addressing challenges related to interpretability and generalizability. Furthermore, this focus area prioritizes development of AI approaches that consider ethical implications, encouraging equitable outcomes across patient populations, and maintaining the highest standards of data privacy and security.