Improving the Detection of Hypertrophic Cardiomyopathy Using Machine Learning Applied to Electronic Health Record Data

PROJECT SUMMARY Hypertrophic cardiomyopathy is the most common inherited cardiac muscle disease with an estimated 750,000 affected individuals in the United States. However, only about 100,000 people have been diagnosed, suggesting that there are significant diagnostic and treatment gaps for individuals with pre-clinical or overt disease, as well as for their at-risk family members. Therefore, it is important to identify individuals who should undergo evaluation for earlier diagnosis and targeted treatment, prior to the development of highly morbid outcomes including heart failure, arrhythmias, stroke, and sudden death. The electronic health record offers a source of high dimensional, longitudinal phenotype information that can be leveraged to create more sensitive and specific diagnostic algorithms. In this patient-oriented mentored career development award proposal, Dr. Nosheen Reza aims to improve the ability to identify individuals with hypertrophic cardiomyopathy through creation and evaluation of machine learning classification models that leverage electronic health record data derived from diverse populations. In Aim 1, she will derive and validate a multi-institutional electrocardiogram-based model for the detection of hypertrophic cardiomyopathy using data from the Penn Medicine electronic health record and will evaluate whether the addition of additional electronic health record-derived traits to this model improves the model's ability to detect patients with hypertrophic cardiomyopathy. In Aim 2, she will externally validate the best performing electronic health record-derived models in two large independent health systems. In Aim 3, she will use implementation science methods to identify clinician-specific barriers to and facilitators of accurate and timely diagnosis of hypertrophic cardiomyopathy and assess clinicians' attitudes toward the use of an electronic health record-derived diagnostic model for hypertrophic cardiomyopathy. Taken together, these aims will lead to prospective dissemination and implementation studies of a generalizable electronic health record-derived diagnostic tool to facilitate early recognition and risk stratification of individuals with hypertrophic cardiomyopathy. Dr. Reza, an early career investigator and genetic and advanced heart failure cardiologist, has a long-term goal of becoming an independently funded cardiovascular data scientist with a focus on applying clinical informatics tools that leverage electronic health record and genomic data to enable precision medicine in the care of patients with cardiomyopathy and heart failure. This K23 award will support Dr. Reza in achieving this goal through a comprehensive and rigorous training plan in bioinformatics, machine learning, and implementation science. Dr. Reza will be supervised by an outstanding mentorship and advisory team at the University of Pennsylvania consisting of national leaders in genetic cardiomyopathies, electronic health record-based research, and translational bioinformatics. The mentored research and career development plan outlined in this proposal will guide Dr. Reza's transition to an independently funded research career.