Improving Cardiology-Rehospitalization Prediction Through the Synergy of Process Mining and Deep Learning: An Innovative Approach

Abstract

Nowadays, hospitals are facing the need for an accurate prediction of rehospitalizations. Rehospitalizations, indeed, represent both a high financial burden for the hospital and a proxy measure of care quality. The current work aims to address such a problem with an innovative approach, by building a Process Mining-Deep Learning model for the prediction of 6-months rehospitalization of patients hospitalized in a Cardiology specialty at San Raffaele Hospital, starting from their medical history contained in the Patients Hospital Records, with the double purpose of supporting resource planning and identifying at-risk patients.