We all have many diseases in our lifetimes. While there are unique genetic and environmental risk factors for each disease and disease subtype, many are shared between diseases. These can contribute to elevated risks for multiple diseases, which may in turn result in specific disease trajectories one will have in one’s lifetime. In this project, we aim to use diagnosis and prescription records from the UK Biobank dataset to characterize these disease pathways and understand their genetic basis. Specifically, we want to
1. Use machine learning approaches to correct potential existing misdiagnoses in medical records, and predict future diseases;
2. Apply existing and develop new statistical methods to understand how genetic pathways leading to two different diseases may lead to their comorbidity; and
3. Develop new statistical methods to identify the physiological measures affected by the genetic risk factors contributing to diseases, and how they may contribute to disease risk
These developments will improve diagnoses and predictions of diseases, as well as better our understanding of diseases that happen in our lifetimes. We hope the genetic findings we make will eventually result in more effective drug targets, leading to more effective treatments of diseases, as well as better preventive measures.