I-Corps: Machine Learning based Clinical Decision Support Tool to Predict 30-day Hospital Readmissions for Congestive Heart Failure Patients

Johns Hopkins University, Baltimore MD

Investigators

Abstract

The broader impact/commercial potential of this I-Corps project is the development of a clinical decision support software tool to predict 30-day hospital readmissions for patients diagnosed with congestive heart failure (CHF). CHF is a chronic cardiac disease where the heart does not adequately pump blood to the body. CHF is known to be the number one reason for hospital admissions and readmissions. Hospital readmissions are independently associated with negative patient outcomes, and Medicare spends  $2.7 billion on CHF readmissions annually. On top of that, an estimated 70% of those readmissions are preventable, and as a result, the Centers for Medicare and Medicaid Services (CMS) started to penalize health organizations for high rates of readmission. Re-admission penalties were up to $550 million across all hospitals in 2018, with ~2,500 hospitals being affected. The software tool may be generalized to other diseases beyond CHF and has the potential to help reduce these avoidable readmissions improving patient outcomes and leading to cost savings for patients, payers, and hospitals. This I-Corps project is based on the development of a clinical decision support tool to predict 30-day hospital readmissions. This is a machine learning-based software tool supported by large-scale clinical datasets to predict the likelihood of readmissions for patients admitted to the hospitals and diagnosed with congestive heart failure (CHF). This technology is a cloud-based software tool that will integrate with electronic health records (EHRs) and can assist case managers with triage and planning of congestive heart failure patients. Building on current information from case managers and quality improvement directors, the team gathers additional information to gain a deeper understanding of the clinical workflow (e.g., attending physicians, transition guides, case managers, directors, and chief nursing officers), revenue streams (e.g., hospital purchasing committees), reimbursement requests (hospitals and payers), and regulatory requirements (e.g., FDA 510(k) regulatory clearance process) to establish connections with potential partners in industry. The goal is to build an evidence-based product solution by interviewing these stakeholders and to establish plans to move the commercialization of the technology forward. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.