Machine Learning–Based Approaches for COVID-19 Transmission Modeling and Prediction: A Comprehensive Investigation

Due to several outbreaks of large infectious diseases, it is crucial to make accurate predictions of dynamics of the pandemics. Among all the tools, machine learning proves to be an efficient way for predicting infectious diseases, such as Covid-19. This paper reviews the use of machine learning algorithms to forecast the transmission dynamics of COVID-19. It outlines the fundamental workflow of both traditional methods (Random Forest, Decision Tree and Linear Regression) and deep learning models (Bi-Gru, RNN, DNN). While these data-driven approaches demonstrate significant capability in capturing complex patterns and making predictions, there are still reliability issues of these methods in real-world health decision-making, such as interpretability, applicability and error accumulation. To address the limitations, the paper proposes three methods, including expert system, transfer learning and detailed data cleansing and collection process. The conclusion advocates for hybrid frameworks that merge professional epidemiological knowledge with machine learning to build more robust and reliable tools for public health decision-making.