An LSTM-based Deep Learning Approach with Application to Predicting Hospital Emergency Department Admissions

Abstract

Since the need for medical cares has significantly increased all over the last years, the efficient management of patient flow becomes a core element for hospitals and particularly in emergency departments (EDs). With the high demand for ED services, overcrowding can be generated and thus the quality of medical services could be degraded. In this regards, forecasting daily attendances at the ED is vital to mitigate the overcrowding problems. Specifically, ED demands forecasting provides relevant information for ED's managers to appropriately use the available resources. This paper presents a Long Short-Term Memory (LSTM)-based deep learning approach for forecasting daily admissions at an ED. Experimental data from the pediatric emergency department at Lille regional hospital center, France, are used to test the efficiency of the proposed approach. Results show the good potential of the LSTM-based deep learning approach in forecasting ED admissions.

Ähnliche Arbeiten