Mathematics is closely related to the theory of machine learning. In this talk, we investigate how mathematical machine learning can help us understand the biomedical and epidemic mechanisms of COVID-19 by three recent studies. First, we look into real-world implications of a rapidly-responsive COVID-19 spread model via deep learning and mathematical modeling. The methodology could also be employed for short-term prediction of COVID-19, which could help the government prepare for a new outbreak. Second, we develop a machine learning model that predicts the mortality of infected patients by using basic patients’ information such as age, residence, comorbidity, and past medical history, which can lead to a medical system that allows patients to check their own severity. Third, we analyzed the difference in the immune response of T cells between COVID-19 patients and non-patients using artificial intelligence. This method was superior to the existing statistical methods in classifying COVID-19 patients and was also effective in predicting the severity of the disease.