The following is a summary of “A survey on diabetes risk prediction using machine learning approaches,” published in the November 2022 issue of Primary care by Firdous, et al.

Chronic diabetes mellitus (DM) can have a number of negative effects. Age, inactivity, a sedentary lifestyle, a family history of diabetes, high blood pressure, stress and depression, poor diet, and other variables can all contribute to the development of diabetes. Diabetes increases a person’s chance of illnesses, including heart disease, stroke, diabetic nephropathy, diabetic retinopathy, nerve damage, and eye difficulties. Around the world, 382 million people have diabetes, according to the International Diabetes Federation. About 592 million people will be living in the country by 2035. Numerous individuals fall prey to it daily, and many are unaware of their vulnerability. People between the ages of 25 and 74 were the main demographic affected. Diabetes can cause a wide range of consequences if it is not properly managed and diagnosed. On the other hand, the introduction of machine learning strategies resolves this significant problem. For a study, researchers sought to examine how machine learning algorithms were utilized to diagnose diabetes mellitus, one of the most dangerous metabolic illnesses today, at an early stage.

Data was gathered from databases including Pubmed, IEEE xplore, and INSPEC, as well as from secondary and primary sources where methods based on machine learning methodologies utilized in healthcare to predict diabetes at an early stage are described.

Following an examination of several research publications, it was discovered that machine learning classification methods, such as Support Vector Machine (SVM), K-Nearest Neighbor (KNN), and Random Forest (RF), etc. exhibited the highest accuracy for early diabetes prediction.

Effective diabetes treatment depends on early identification. Many people were unaware of their level of it. In the study, the entire evaluation of machine learning techniques for early diabetes prediction and the application of several supervised & unsupervised machine learning algorithms to the dataset were covered. To develop a more accurate and comprehensive predictive model for diabetes risk prediction at an early stage, the study would also be built upon and improved. It was possible to accurately diagnose diabetes and evaluate performance using a variety of indicators.