The prevalence of diabetes has become one of India's most pressing health concerns. Hyperglycemia refers to a group of syndromes characterised by elevated blood sugar levels. It's a chronic illness that interferes with the body's normal blood sugar management mechanisms. The field of medical sciences is seeing a rise in interest in diabetes mellitus prevention and prediction. This research seeks to survey the many methods currently in use for diabetes prediction. In this paper, we examine the research of several writers on diabetes prediction techniques. The purpose of our research into diabetes prediction models was to identify criteria for vetting studies and compiling relevant findings. It is difficult to analyse diabetic data since most medical data are nonlinear, nonnormal, correlation organised, and complex. Algorithms based on machine learning are not allowed to be used in the medical and healthcare industries. Predicting diabetes mellitus early requires a method that is distinct from the methods now in use. Patients can be classified as diabetes or non-diabetic using a risk stratification technique based on machine learning. Our research is highly recommended because it draws from a variety of articles that will aid other academics working on different diabetic prediction models.