A Survey on Diabetes Detection Based on Machine Learning Classifiers

Computer-assisted sickness diagnosis is less costly, time-saving, accurate and it eliminates the need for extra employees in medical decision-making. According to many nutrition studies, nearly a quarter of the world's population suffers from chronic illnesses such as diabetes. As a result, an effective machine learning regressor capable of correctly diagnosing diabetes is urgently needed. The aim is to determine whether individual classifiers or a set of classifier combinations provide the best diabetes detection accuracy. The concentration of glucose in human blood is known as the blood glucose level. Diabetes is a long-term condition marked by a lack of glucose. To maintain blood sugar levels within the suggested target range, insulin therapy is necessary. As per the statistic from the WHO, diabetes affects almost 400 million individuals worldwide. The need for regular blood glucose monitoring in the therapy phase cannot be overstated. Diabetes can be detected in a variety of ways. ML models are utilized to determine whether or not the patient has diabetes to confirm the sickness. This paper attempts to survey various recently published literature that have proposed different approaches to detecting Diabetes Mellitus using machine learning techniques. Some of the surveyed techniques utilise comparative analysis of ML as well as DL classifiers, while others techniques use medical records to identify the risk predictor.