1. On the Evolution of COVID-19 Virus Based on the Prediction Model of Deep Learning and Emotion Analysis.
- Author
-
Yucheng Fan and Dong Qiu
- Subjects
- *
DEEP learning , *SARS-CoV-2 Omicron variant , *PREDICTION models , *HEALTH boards , *COVID-19 , *SENTIMENT analysis - Abstract
In January 2020, COVID-19 broke out in Wuhan, China. In just a few months, the virus spread around the world. In the past three years, the virus has undergone continuous mutations, with the latest variant being the Omicron variant. Whenever a new variant emerges, there are significant changes in the transmission rate, mortality rate, and other essential disease characteristics. These characteristics of the virus have posed considerable challenges to countries and health departments. Time series models used in this study incorporate text sentiment. To achieve this, we utilized snscrape to retrieve tweets and applied different keywords to filter the tweets. Subsequently, we used clean text as input for a pre-trained model to conduct the sentiment analysis. Finally, the sentiment analysis results and other epidemic features were combined as inputs for time series models to generate predictions. We can observe changes in the virulence of different variants through the models' outcomes. Research has shown that incorporating sentiment analysis results can effectively improve the model's predictive performance. When a model is trained using historical data, it cannot accurately predict viruses that will emerge in the future. The same holds in reverse. The result indicates that the nature of the virus has changed during different stages, suggesting the emergence of new variants. The study will help local health departments improve control measures, enabling adjustments to be made specifically for different variants. [ABSTRACT FROM AUTHOR]
- Published
- 2024