Application of a SARIMA model in predicting varicella incidence in Chongqing

Wang Ju; Li Baisong; Peng Yang; Xiong Yu; Yang Jule; Li Zhijin; Qi Li; Long Jiang

doi:10.11986/j.issn.1673-873X.2025.04.06

Abstract：Objective To explore the application of the Seasonal Autoregressive Integrated Moving Average(SARIMA)model in predicting varicella incidence trends in Chongqing based on surveillance data.Methods Varicella case data and population data from January 2015 to December 2024 were retrieved from the Infectious Disease Reporting Information Management Sys-tem.R software was used to construct the optimal SARIMA model for monthly varicella incidence in Chongqing.The model was used to predict the reported incidence of varicella from January to December 2024,with predictions compared against actual values.Model fit was evaluated using mean absolute error and other metrics.Varicella incidence for January to December 2025 was forecasted.Results From 2015 to 2019,reported varicella incidence in Chongqing showed an upward trend,while fluctua-ting changes with an overall downward trend occurred from 2020 to 2024.The SARIMA(1,0,1)(0,1,2)12 model was optimal,with goodness-of-fit R2=0.896 and Bayesian Information Criterion=475.356.Residuals constituted a white noise series,and all model parameters passed statistical tests.For predictions from January to December 2024,the mean absolute error between actual and fitted values was 1.182,the root mean square error was 1.660,and the mean percentage error was-1.263%,mean absolute percetage error was 23.841%indicating reasonable model performance.Conclusion The constructed SARIMA(1,0,1)(0,1,2)12 model demonstrates satisfactory fit and can be used for short-term prediction of varicella incidence trends in Chongqing,providing certain guiding implications for developing varicella prevention and control measures.