Forecasting of weekly patient visits to emergency department: real case study

Abstract

Emergency department (ED) is the most crowded entity in hospitals, because it is the access point of almost all patients looking for care without beforehand appointment. Accordingly, accurate forecasts of ED visits is increasingly required to bring up ED throughput. Hence, combining Artificial Neural Networks (ANNs) with a signal decomposition technique named Ensemble Empirical Mode decomposition (EEMD), to make one-ahead forecasting of patients arrivalstoED, is newly investigated in this paper. Seven years of aggregated weekly demand, from 2010 to 2016, has been collected from all services of emergency department of the University Hospital Hassan II of Fez city of Morocco. The time series (TS) of the demand was decomposed into several sub-signals, each of them was modeled using an ANN model. Then, their forecasting results were combined to produce the total forecast. Finally, the results of the used model were compared against the benchmarking models:ANN without signal decomposition, ANN with Discrete wavelet Transform (DWT) decomposition and ARIMA model. The results of this investigation show that, in forecasting ED weekly visits, ANN assisted with EEMD outperforms the benchmarking models for approximation and generalization capabilities, while overcoming the problem of overfitting. Thus, the used model can be employed to forecast efficiently ED arrivals, and to optimize human and material resources of hospitals.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 46 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.