Epidemiological Forecasting Models Using ARIMA, SARIMA, and Holt–Winter Multiplicative Approach for Pakistan

Abstract

Background of the Study. Statistical models have been extensively used in modeling and forecasting the different fields of agriculture, economics, social sciences, and medical sciences. The transmission of some diseases is a serious life threat around the globe; therefore, proper assessment and modeling need time. Malaria is one of the major life-threatening diseases in Pakistan, and some death cases due to this disease have been reported during the last decade. Methodology. The data have been collected from the Ministry of Health, Rahim Yar Khan, Pakistan, from January 2011 to March 2022. Data were analyzed by applying time series models for prediction purposes. Diagnostic measures such as RMSE, MAE, and MAPE were used to choose the best forecasting model. Results and Discussion. This study aims to forecast malaria cases by choosing the best forecast model. After comparison, it was concluded that the Holt–Winter multiplicative model outperformed the ARIMA and SARIMA models, with the lowest RMSE, MAPE, and MAE compared to other models. Malaria cases in the district Rahim Yar Khan were forecasted by the Holt–Winter multiplicative model, for the month of April 2022 to January 2023. From the forecasting results, the minimum number of cases was found to be 586.75 in June 2022 and the maximum number of cases was found to be 1281.93 in October 2022 among the next ten months. Based on the results, it is paramount for the GOP (Govt. of Pakistan) to enhance the vaccination policy to erase the impacts of malaria cases to flatten the curve.