[Research on the prediction of Hepatitis C incidence trend in Taiyuan City based on combination model].

Objective: Based on the autoregressive integrated moving average (ARIMA) model, back propagation neutral network (BPNN), and ARIMA-BPNN model, select the optimal model suitable for predicting the incidence trend of hepatitis C in Taiyuan City according to the characteristics of the data. Methods: The data of reported cases of hepatitis C in Taiyuan from 2008 to 2021 were selected, and the seasonal trend decomposition chart was used to analyze the seasonal characteristics of the monthly incidence rate of hepatitis C in Taiyuan during the period, and the ARIMA model, BPNN model, and ARIMA-BPNN model were established to predict. The performance of the model was measured using four indicators: mean absolute error (MAE), mean squared error (MSE), root mean square error (RMSE), and mean absolute percentage error (MAPE). Results: A total of 20 025 cases of hepatitis C were reported, and the overall incidence trend was stable. The BPNN model performed well on MSE, MAE, and RMSE indicators, the ARIMA-BPNN model performed well on MAPE indicators, and the ARIMA model performed relatively averagely. Conclusions: The ARIMA-BPNN model is a better model for predicting the trend of hepatitis C in Taiyuan City, with a higher predictive performance than a single model. It has significant prospects in predicting the trend of infectious diseases.