Patient Visits Forecasting Based On Decomposition And Ensemble Paradigm

Variation of patient visits to a hospital plays an important role in hospital running,for instance,staffing,disposable resource arrangement as well as policy and decision making etc.depend on patient visits to a hospital.Obviously,the hospital operation benefits from accurate patient visits forecasting.To improve the accuracy of predicting visits to hospitals,in the principle of "decomposition and ensemble",a decomposition-and-ensemble approach based on wavelet decomposition(WD)is proposed.In particular,wavelet decomposition(WD),the most popular date decomposition tool,is employed to decompose the original data of patient visits to hospitals into a set of meaningful components,which largely simplifies the complex data and reduces the prediction difficulty.Second,the most typical artificial intelligence(AI)model,i.e.,artificial neural network(ANN),is implemented to model each decomposed component.Third,all individual prediction results are aggregated into the final prediction output.For illustration and verification,four monthly series data of patient visits to Chinese hospitals are introduced as the sample data,and the results indicate that the proposed WD-based model can generate much more satisfactory results than all considered popular forecasting techniques,in terms of prediction accuracy.The proposed WD-based decomposition-and-ensemble model is capable of tackling the vitality and complexity of patient visits data and forecasts patient visits to hospitals with higher precision.It is implied that the proposed WD-based learning paradigm can be used as one promising prediction tool for patient visits to hospitals.