Exploring the Efficacy of Machine Learning Algorithms for Diabetes Prediction: A Comparative Prediction

Abstract

Abstract: We know that diabetes is one such disease that affects millions of people worldwide, so its early detection and accurate prediction can help in the timely intervention and management of the disease. The main objective of this project is to do a comparative analysis of the performances of the different machine learning algorithms like Random Forest, Support Vector Machine(SVM), K-Nearest Neighbors(KNN), and a Hybrid Random Forest in predicting diabetes and for this the various patient attributes like age, BMI, glucose level, blood pressure, etc are included in the dataset for training and testing the models and various evaluation metrics like accuracy, precision, recall, and F1-score of the different algorithms are used to do the comparative analysis. The Hybrid Random Forest algorithm is found to outperform the other considered algorithms with an accuracy of 90.4%. Feature selection is also involved in the study to identify the most important variables that would help in effective prediction of diabetes and the overall analysis demonstrates that glucose level, BMI, and age are the top variables that are considered to be very important for diabetes prediction. So in healthcare, identifying the risk of developing diabetes at an earlier stage and taking measures for its prevention and helping many patients can be a very beneficial findings of this study