Survey paper on machine learning algorithms for diabetes prediction

Abstract

A long-standing ill health which straight away attacks the pancreas, and the body can’t produce insulin is diabetes. Insulin is principally important to perpetuate the level of blood glucose. Diabetes can be caused by patients due to a lot of characteristics like the uncontrolled weight of the body, lack of physical exercise, high blood pressure, and unusual cholesterol levels. A lot of complicatedness can be caused, however, one of the most common ones is an increase in urination. Diabetes can harm the skin, nerves, and eyes. If left untreated, it can also result in kidney failure and the eye condition diabetic retinopathy. 537 million people worldwide had diabetes in 2021, according to data from the International Diabetes Federation. Based on 2019 statistics, 7.10 million persons in Bangladesh have contracted this disease. The majority of the population is affected by diabetes, which is the main disease increasing the ratios and leading to renal failure, amputations, blindness, lower limb amputations, stroke, and heart issues. Numerous factors, including a lifestyle devoid of regular exercise, eating unhealthy foods, being overweight, heredity, and more, could be the cause. Foods are converted into glucose by the human body. A higher blood glucose level is used to categorize a group of diseases known as diabetes. In individuals with diabetes, the pancreas is unable to produce insulin. In this case, machine learning techniques are crucial for forecasting this illness. As a supervised machine learning technique, classification is the best-preferred methodology for labeled data categorization. The medical field uses a variety of machine-learning approaches to anticipate and identify illnesses. The main aim of this paper is to compare various algorithms of machine learning for diabetes prediction.