Abstract 16941: Machine Learning Enhanced Predictions of Hospital Readmission or Death in Heart Failure

Introduction: Readmissions are common, costly and often preventable. The LACE risk score is an established index to quantify the risk of readmission or death. We used machine learning to develop a Heart Failure (HF) specific predictive tool. Methods: The OM1™ Cardiology data warehouse contains deep clinical and claims data on patients seen in cardiology practices across the US. Patients with HF, hospitalized between Oct 2014 and Sept 2016, with at least 12 months of data before the index admission, and 30 days of data post discharge, were included. The unit of analysis was hospitalization; those occurring before Apr 2016 (~70%) were used as the training set and the remainder as the validation set. Predictive features were developed by machine learning for the training set, and the performance of the resultant risk score was compared with that of the LACE risk score for the validation set. Results: The study included 14,065 HF related hospitalizations with 3,502 (25%) unplanned readmissions or death within...