Forecasting the incidence trend of varicella in Haidian District,Beijing based on Prophet and NeuralProphet models
WEI Yiyun
SUN Yamin
LIU Xuanzhuo
DU Jing
Abstract：Objective To investigate the epidemiological trends of varicella in Haidian District,Beijing,using Prophet and NeuralProphet(NP)models,and to provide evidence-based insights for optimizing varicella control strategies.Methods The weekly varicella cases data in Haidian District from Week 1 of 2009 to Week 26 of 2024 were analyzed.The Prophet and NP models were trained on data from 2009 to 2023,with hyperparameters optimized via the Optuna algorithm.Model performance was evaluated on the 2024 test set(26 weeks)using root mean squared error(RMSE),mean absolute error(MAE),and mean absolute percentage error(MAPE).Model components were decomposed to identify contributing factors.Results Two annual incidence peaks of varicella were observed in Haidian District.The incidence of varicella exhibited a continuous decline over the years,while the autoregressive component within the model demonstrated a progressive attenuation of fluctuations starting from 2012.The Prophet model yielded RMSE,MAE,and MAPE values of 9.489,7.936 and 27.408%,respectively,while the corresponding metrics for the NP model were 6.102,4.848 and 18.190%.Conclusions The Prophet model shows moderate applicability for predicting varicella trends,whereas the NP model improves forecasting accuracy.By analyzing the components of the model,scientific evidence and data support can be provided for evaluating the effectiveness of measures,allocating resources rationally,and formulating effective prevention and control strategies.
Keywords：Prophet modelNeuralProphet modelvaricellaforecast
Publication Date：2025-06-30
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)
Pages：5( 268-272 )