Developing a Diabetes Prediction Decision Support System with Machine Learning

Chronic diseases like diabetes pose a major threat to global healthcare, with estimates suggesting an increase from 382 to 1 million cases in 2013 to 692 million by 2050. Characterized by high blood sugar and symptoms such as increased thirst and hunger, diabetes complications occur if a person has this ailment and does not take care of her health properly. Normally, our body converts food into fuel (glucose), and the pancreas produces insulin, a hormone that unlocks cells and allows glucose entry. There are different types of diabetes of which the main categories are type 1 and type 2. To improve early detection, scientists are using machine learning, a field that allows machines to learn from experience. This project explored the combination of various machine learning algorithms for more precise identification of early diabetic symptoms. After evaluating the accuracy of various algorithms, including Gaussian Naive Bayes 96.43%, Support Vector Machine 96.69%, Linear Regression 98.52%, and Decision tree 98.86%, we have chosen the decision tree algorithm for its superior performance.