Project Summary
Adult-onset Type 2 Diabetes Mellitus (T2DM) affects 37.3 million individuals in the United States. The combined treatment and reduced productivity costs associated with T2DM are $327 billion per year in the United States alone. Importantly, T2DM disproportionately impacts non-Hispanic Black/African American and Hispanic/Latino individuals compared to Asian/Asian American and non-Hispanic White/European American individuals, leading to increased disease burden and health disparities in historically under-resourced populations. These important differences in T2DM and its comorbidities indicate race/ethnicity and/or genetic ancestry can somehow affect disease risk. To date the majority of genetic research has focused on European descent individuals, leaving the genetic contribution to T2DM risk poorly understood outside the European context.
Recent studies have demonstrated that T2DM could be further classified by the long-term comorbidities patients develop. This has the potential to inform what comorbidities patients are most at risk for and provide a potential opportunity to provide preventative medicine.
The purpose of our project is to determine specific T2DM clusters associated with long-term T2DM outcomes, utilizing unsupervised machine learning methodologies. Our long-term goal is to create electronic medical record alerts that predict each patient's T2DM subtype and inform physicians of the potential complications of diabetes their patients are most at risk for. Collectively, this could improve personalized medicine, decrease racial and ethnic disparities in diabetes comorbidities, and decrease comorbidities and complications, all while saving money and improving patient quality of life.