Application of optimized combination prediction model in the prediction of hand, foot and mouth disease
10.3969/j.issn.1006-2483.2026.01.012
- VernacularTitle:优化组合预测模型在手足口病发病预测中的应用
- Author:
Weijie TIAN
1
;
Qian GAO
2
;
Kun YANG
2
;
Zhirong ZHAO
2
;
Jian CHEN
3
Author Information
1. Wannan Medical College, Wuhu, Anhui 241000, China
2. Ma'anshan Center for Disease Control and Prevention, Ma'anshan, Anhui 243000, China
3. Wannan Medical College, Wuhu, Anhui 241000, China;Ma'anshan Center for Disease Control and Prevention, Ma'anshan, Anhui 243000, China
- Publication Type:Journal Article
- Keywords:
Hand, foot and mouth disease;
Auto-regressive integrated moving average model;
Singular spectrum analysis;
Long short-term memory model
- From:
Journal of Public Health and Preventive Medicine
2026;37(1):58-62
- CountryChina
- Language:Chinese
-
Abstract:
Objective To explore scientific and accurate prediction methods for the incidence of hand, foot, and mouth disease in the post-pandemic era, and to address modeling challenges caused by abnormal fluctuations in case numbers from 2020 to 2023. Methods The seasonal index was used to pre-process the data. The traditional seasonal autoregressive integrated moving average (SARIMA) model, singular spectrum analysis (SSA)-ARIMA model, ARIMA-Long short-term memory (LSTM) model, and SSA-ARIMA-LSTM model were used to fit the incidence from 2013 to 2023, and the incidence of hand, foot and mouth disease in 2024 was predicted. The real data collected in 2024 were used as the test set to compare the prediction performance of the models. Results The fitting performance of the constructed models was as follows: the ARIMA model had MAE=107.50 and RMSE=144.53, the SSA-ARIMA model showed MAE=2.84 and RMSE=4.33, the ARIMA-LSTM model achieved MAE=99.46 and RMSE=131.59, and the SSA-ARIMA-LSTM model had MAE=96.35 and RMSE=132.13. In terms of prediction performance, the ARIMA model resulted in MAE=151.64 and RMSE=146.70, the SSA-ARIMA model demonstrated MAE=41.22 and RMSE=57.01, the ARIMA-LSTM model yielded MAE=220.75 and RMSE=257.89, and the SSA-ARIMA-LSTM model recorded MAE=58.83 and RMSE=72.06. Conclusion The SSA-ARIMA model has the best fitting degree and the highest prediction accuracy, and is suitable for predicting the incidence trend of hand, foot and mouth disease.
