Prediction of complex disease occurrence and progression with polygenic risk scores, machine learning, and multimodal data.
Approved Research ID: 88417
Approval date: September 14th 2022
The intent of this proposal pertains to improving our ability to correctly diagnose distinct diseases that may manifest in similar ways (e.g., bipolar disorder and major depressive disorder), or to suggest a diagnostic test for at-risk participants that show few detectable signs of disease, such as with atrial fibrillation. We expect the proposed research to take at least a year to complete. Because disease is a combination of genetic and non-genetic factors (e.g., locale where someone lives; socioeconomic status), our approach involves developing predictive computer models that learn how to combine many different types of common measurements with a form of genetic risk assessment (polygenic risk score), a recent development in clinical medical genetics. To ensure the broad applicability and usefulness of these models, we place special emphasis on ensuring that they are applicable to populations of different genetic ancestries. We believe that our approach will substantially improve our ability to more correctly diagnose diseases and improve their treatment by enabling earlier actions that are tailored to the individual.