Investigation of factors increasing waiting times in the Emergency Departments of “San Giovanni di Dio e Ruggi d'Aragona” Hospital through machine learning

Abstract

Emergency Departments (EDs) overcrowding is an acknowledged critical issue affecting international public health in recent years, that arises from both the growth of the health care supply/demand imbalance and the lack of beds available in hospitals wards and EDs. Emergency department length of stay (ED-LOS) is identified as a valuable key measure of EDs bottlenecks, and specifically of the rapidity of access to care for patients and of the overcrowding. ED-LOS measures how long patients stay in the ED from their first registration and triage to their admittance to a hospital ward or their discharge. Prolonged ED-LOS has been associated with adverse outcomes, such as reduced level of quality of care and patient satisfaction, increased risk of mortality and financial loss. Understanding aspects affecting LOS is essential for the management of an ED and for implementing improvement interventions. The aim of this study is to determine the several factors affecting LOS in EDs and to build a model capable of predicting ED-LOS through different machine learning (ML) models. ML algorithms were performed considering data extracted from the ED database of the “San Giovanni di Dio e Ruggi d'Aragona” University Hospital (Salerno, Italy). The proposed prediction model shows promising outcomes and therefore it can be used for the prediction and governance of the ED-LOS, thus anticipating the occurrence of overcrowding and improving ED care and efficiency.