Principal Investigator: Dr Christopher Haggerty
Geisinger Clinic, Danville, Pennsylvania, USATags: 55725, genomics, heart disease, Imaging, Machine Learning
Heart disease is the leading cause of death worldwide. In order to effectively diagnose and treat patients with heart disease, physicians must combine and mentally process data from different sources, including imaging, electrocardiograms (ECG), family history, genomics, and more. Additionally, many of these data sources contain more information than a physician can see, but which computers can quickly and easily measure. Large collections of health data such as the UK BioBank have recently emerged and shown promise to help ease this burden and improve patient outcomes.
The goal of our research is to develop computer models of heart disease that can process multiple data types and provide valuable insight specific to each individual patient. We have developed similar models using data at our institution, but we need to make sure they are consistent for different data sources and groups of people. Firstly, we will develop models that calculate a patient’s risk for disease based on their genetic profile. Second, we will develop models to automatically extract key information from complex data, such as the volume of the heart from medical images or the presence of irregular heart rhythms from an ECG. Finally, we will combine these two model types, to help the physician make better informed decisions regarding patient diagnoses and treatment.