Prediction of COVID-19 spread via LSTM and the deterministic SEIR model

Due to the outbreak of COVID-19, China and most countries in the world have been seriously affected and tens of thousands of people have lost their lives, it is urgent to study the transmission characteristics and trends of the virus. In this study, we adopt the Long Short Term Memory algorithm at first to predict the infected population in China. However, it does not explain the dynamics of diffusion process, and the long-term prediction error is too large. Therefore, the widely-accepted SEIR model is introduced to capture the spread process of COVID-19. By using a sliding window method, we suggest that the parameter estimation and the prediction of the infected populations are well performed. This may provide some insights for epidemiological studies and understanding of the spread of the current COVID-19.