Application of the ARIMA Model in Predicting Disease Incidence

Abstract

The topic of this review paper is the application of the regressive Integrated Moving Average (ARIMA) model in disease prediction. As one of the most popular models in time series prediction, ARIMA is widely used and receives extensive attention. Predicting the incidence of diseases is important, as it can help the public make timely strategies or prepare in advance for the future, such as the number of medical facilities. This article presents successful cases of ARIMA and models that outperform ARIMA. The conclusion is that the ARIMA model can indeed provide reliable prediction results in many situations. However, when dealing with data that has complex nonlinear characteristics or high volatility (such as the incidence rate of certain infectious diseases), more advanced machine learning models (such as LSTM and random forest) or traditional yet adaptable models (such as Holt-Winters) often exhibit lower prediction errors and stronger fitting capabilities. This article holds that the future direction of scientific research should not be limited to the application of a single model but rather should actively explore the "hybrid model" or "integrated model" that combines the advantages of the ARIMA model with those of other models.