Dự báo lượng bệnh nhân ngoại trú tại bệnh viện sử dụng mô hình Holt-Winters, mạng nơ ron nhân tạo ANN và LSTM

Abstract

Effective hospital outpatient forecasting is an important task for mordern hospitals to implement intelligent management of medical resources. As outpatient visits flow may be nonlinear, seasonal and dynamic, we explore to compare empirically two typical forecasting methods: Holt-Winters and artificial neural network (ANN) in order to identify the best method in this challenging forecasting problem. Holt-Winters belongs to statistical analysis category while ANN is a machine learning method. We applied the two methods to forecast daily outpatient visits flow at General Hospital of Cu Chi Area in Ho Chi Minh City. Experimental results reveal that Holt-Winters outperforms ANN in the hospital outpatient visits flow forecasting. The MAPE indicator of Holt-Winters is 12.25%.