Predictive and early warning analysis of infectious gastroenteritis based on the BiLSTM-BiGRU model.

Appropriate use of scientific early-warning infectious disease surveillance methods plays a vital role in disease control and prevention. Recently infectious gastroenteritis has become an important public health problem. In consideration of meteorological factors strongly linked with the incidence of infectious gastroenteritis, we obtained data on the number of infectious gastroenteritis cases and meteorological data from January 2008 to June 2023, a total of 808 weeks. We constructed a BiLSTM-BiGRU model to fit and predict the incidence of infectious gastroenteritis in Tokyo, Japan, to improve the prediction accuracy and early warning efficiency of infectious gastroenteritis, provide references for relevant departments to formulate infectious disease prevention and control measures in advance, and make emergency preparations. For this purpose, we also used three optimization algorithms for parameter tuning and constructed a moving percentile control chart warning model. The results show that the BiLSTM-BiGRU model performed better than mainstream deep learning methods. Among the three selected optimization algorithms, the Grey Wolf Optimization algorithm performed the best, with an R of 0.85, and led to reductions of 11.90 % in RMSE, 12.44 % in MAE, and 16.18 % in MAPE, respectively. We found that the GWO-BiLSTM-BiGRU model can fit and predict the number of infectious gastroenteritis cases in Tokyo accurately. Relevant departments should be alert to the high incidence of infectious gastroenteritis during weeks 3-5 each year based on the prediction and warning results.