Abstract: Delirium is one of the common complications of hospitalized patients, which seriously affects the prognosis and outcome. Therefore, it is very important for the early prevention of delirium. The premise of prevention is to effectively identify high-risk patients who may have delirium. This paper mainly introduces the application status of machine learning in delirium prevention, so as to promote the construction of delirium prevention and evaluation system, and provide reference for formulating effective delirium prevention strategies, then reduce the occurrence of delirium.