Applications of ML/DL for Predicting Epidemic Outbreaks

Abstract

Officials around the world are using many COVID-19 outbreak prediction models to make intelligent conclusions and implement applicable precautions. Easy epidemiological and statistical models are among the typical prototypes for COVID-19 worldwide pandemic forecasting. Authorities have paid more attention to them, and they are well known in the media. Thus, a high-level mathematical framework has low deep predictive performance due to speculation and insufficient data. Existing models are not generally sufficient in addressing the problem; they ought to be developed further and their accuracy and generality increased. In this article, two deep learning techniques, referred are contrasted with a previous soft method in order to evaluate their performance in predicting a new technology, the "co-variate" and sensitive effect" and "imputed residual incident rate" versions. Out of the multi-level perception models (MLP) and the numerous models researched, only two (complex neural and adaptive particle swarm optimization) highlights the potential. Based on the findings presented here, because of the COVID-19 outbreak's highly dynamic existence and wide range of actions. This study indicates that machine learning can be used to model the outbreak from country to country. This paper serves as an initial benchmarking exercise to show how machine learning can be used in future research. The paper also suggests that by combining machine learning and SEIR models, true novelty in outbreak prediction can be realised.