Introducing an appropriate forecasting system to estimate the treatment demand at Imam Reza hospital of Urmia

The hospitals are responsible for maintaining the health of the population and devote himself a large part of health expenditure. Evidence shows that there is a vast prospect to improve and promote hospitals resources (financial and human). Awareness of the amount of future demand highly ensures optimal management of resources and quality of services in the health field. The aim of this study is to investigate the linear (ARIMA) and nonlinear (neural network MLP) models to estimate the treatment demand at Imam Reza hospital of Urmia, using data at different time intervals such as daily, weekly and monthly in different sections of the hospital. The results of this study indicate that ANN MLP nonlinear model has a good performance in predicting treatment demand and it is capable to provide more accurate forecasts than ARIMA model. Neural network MLP model with an average error of 24.96% compared to ARIMA model with an average error of 26.73% has a high predictive power. Also the estimation results of pediatric ward indicate that the ARIMA model compared to Neural network MLP model has more predictive power, the reason for this inconsistency with research hypotheses must be sought in low variance of this section.