1. Prediction models of COVID-19 based on SEIR and LSTM RNN.
- Author
-
Kang, Yi, Xiao, Jun, Ke, Xiaotian, and Yang, Lv
- Subjects
COVID-19 ,PREDICTION models ,DIFFERENTIAL equations ,MACHINE learning ,EPIDEMICS - Abstract
As the infectious population proliferates dramatically in the past months, the COVID-19 has become a severe global challenge. For the purpose of forecasting the inclination of the epidemic and evaluate the effectiveness of public interventions, we focus on constructing models consisting of differential equations and machine learning in this paper. Applying the SIR and SEIR epidemic models and the LSTM RNN model in Beijing, China, we demonstrate the comparison between models implemented public health interventions and without it. The similar accuracy of SEIR-2 (adopted interventions) and the LSTM RNN model to the true value present the necessity to implement public health interventions such as quarantines to postpone the crest point, obtaining a prolonged time to confront the epidemic. The utility of the model is compelling and worthy to apply in other cities. [ABSTRACT FROM AUTHOR]
- Published
- 2020
- Full Text
- View/download PDF