An Epidemic Trend Prediction Model with Multi-source Auxiliary Data

The global outbreak of epidemics profoundly affects public health and societal development. The development of epidemic trend prediction models is crucial to prevent the recurrence of pandemics. Therefore, we propose a Bayes-Attention AL-Forecast prediction (BALF) model to forecast the future development trends of epidemics. Firstly, we introduce an attention mechanism to integrate the population mobility data features into the case data. Subsequently, based on fused data, we employ an ARIMA-LSTM Forecast (AL-Forecast) model to predict the development trends of epidemics. Finally, experiments are conducted based on real datasets. The results indicate a close correlation between predicted and actual case numbers, and the model’s prediction performance excels with baseline and other state-of-the-art methods. We release our source code at https://github.com/Bevan-Wang/MEHP .