Interpretable Deep Learning Model for Longitudinal Electronic Health Records and Applications to Heart Failure Prediction
Georgia Institute Of Technology, Atlanta GA
Investigators
Linked publications & trials
Abstract
PROJECT SUMMARY Heart failure (HF) is a highly disabling and costly disease with a high mortality rate. In the prediagnostic phase (i.e., 1236 months before diagnosis), HF is difficult to detect given the insidious signs and symptoms. After diagnosis, where it is not possible to reverse disease progression, efforts are made to avoid hospital admission and readmission, but with limited capabilities to stratify patients by risk. We propose to develop interpretable deep learning models applied to largescale electronic health record (EHR) data to detect HF related events on two different time scales. One set of models will be developed to detect HF diagnosis one to two years before actual documented diagnosis. Separately, we propose to identify HF patients who are at risk of hospital admission and readmission . The project focuses on developing deep learning models that offer the potential for greater accuracy, clinical interpretability, and utility than alternatives. The expected deliverables include comprehensive software for creating deep learning algorithms that predict HF outcomes and related software tools for model visualization.
View original record on NIH RePORTER →