Abstract: Electronic Health Records (EHRs) provide a wealth of information from which to conduct research. Complement this data with Medicare Claims and you have a powerhouse of information that tells a patient’s story. We will discuss our early experiences and the technical design for a Machine Learning project working on this data predicting patient Admissions, and its high-performance implementation within a typical enterprise IT environment.
Learning Objective 1: To understand the benefits and challenges in combining claims and EHR data.
Learning Objective 2 (Optional): To learn about a successful data pipeline enabling machine learning on a combined EHR and claims dataset.
Ursula Rogers (Presenter)
Shelley Rusincovitch, Duke University
Erich Huang, Duke
Ricardo Henao, Duke University
Michael Gao, Duke University
Mary Schilder, Duke Health
Eugenie Komives, Duke Health