Machine Learning Models for Predicting and Managing Pandemic Outbreaks: Lessons from COVID-19

Abstract

In this pandemic era, the COVID-19 situation reaffirmed the need for accurate predictive tools that can forecast the dynamism of outbreaks and guide initiatives for timely public health responses. Over the course of multiple pandemics, there has been a growing interest and evidence of the utility of machine learning (ML) models in predicting the shifting trends for infections, the allocation of resources and risk stratification. This review summarizes major ML methods employed during COVID-19, an evaluation of models performance on common publicly available datasets, and sharing of findings on learning for better preparation for future pandemic outbreak. Conclusion: Ensemble and deep learning models are superior to conventional approaches for predicting outbreak trajectories, while incorporations with near real-time data sources improve responsiveness. Also discusses the challenges of data quality + reliance on machine learning, model interpretability, and other ethical considerations. COVID-19 Prediction and Forecasting Model in Pandemic Prediction: A Machine Learning Perspective