Using Data Science to Predict Readmissions in Heart Failure

Abstract

This review describes the current literature on the use of data science to predict readmissions of patients with heart failure. We examine the chronology of heart failure management from the emergency department, inpatient unit, transition of care, and home care. We examine the software and hardware which may improve readmission rates of this common and complex disease process. There are multiple novel applications of data science which have been used to predict readmissions of heart failure patients. In the emergency department, efforts are focused on identifying patients who can be safely discharged after a brief period of stabilization; while inpatient endeavors have attempted to predict those patients at risk for decline after discharge. Overall, prediction rules have had mixed results. Outpatient telemonitoring with invasive devices seems to hold promise. New technologies may be the key to future improvements in readmission rates. Heart failure holds a high morbidity and mortality, and hospitalizations are common. A number of technological interventions have been developed to prevent readmissions in this complex population. Improvements in technology may lead to reductions in heart failure admissions, reduced mortality, and improved quality of care.