Prediction of COVID-19 Confirmed Cases Combining The LSTM Model and Evolutionary Strategy

ABSTRACT

In the global response to the COVID-19 epidemic, a reasonable prediction of the number of infections is a significant reference to reveal the trend of the outbreak and help governments take appropriate action. In this paper, we propose a new ES-LSTM model to predict the growth rate of the number of new infections per day and use a feature processor to address interventions in time series to quantify the impact of interventions to slow the spread of the outbreak. The evolutionary strategy is used to handle the problem that different interventions have different impacts on outbreak prevention and control, as well as optimize model weight to improve the accuracy of prediction results. Experimental results demonstrate that compared to the Linear model, CNN model, and the LSTM model, the MAE of the algorithm is enhanced by 72.9%, 27.6%, and 26.3%, and the RMSE is improved by 74.15%, 31.4%, and 29.5% respectively. © 2022 IEEE.