A Hybrid Prediction Model Based on Decomposition-Integration for Foodborne Disease Risks

Foodborne diseases (FBDs) are contagious, explosive, clustered diseases caused by the ingestion of contaminated foods, which represent huge economic and health burdens globally. Reliably predicting the risk trend of FBDs has become a major challenge in the field of public health. This study aimed to design a risk prediction model suitable for predicting FBD risks by using the decomposition-integration technique. A total of 28,646 FBD cases from FBD surveillance data reported by all sentinel hospitals in Wuxi from 2019 to 2023 were included in the study. The obtained FBD risk data were decomposed into multiple intrinsic mode functions (IMFs) using complete ensemble empirical mode decomposition with adaptive noise, which were then reconstructed by calculating the sample entropy. Finally, the time dependence of the reconstructed IMFs was explored using a temporal convolution network-long short-term memory (TCN-LSTM) model to obtain the prediction results of each component, which were then linearly added to obtain the final prediction results. Compared with other models, our proposed prediction model significantly improved the prediction accuracy of FBD risks, with a best average root mean square error of 5.349 and mean absolute error of 3.819, demonstrating at least a 40% improvement in accuracy over standalone LSTM. The FBD risk prediction results obtained by the method proposed in this study can provide data support for food safety management and policy making and enable more accurate early warning of FBDs.