Early Prediction of Diabetes Using Machine Learning Algorithms

Abstract

We say health is wealth. However, to earn the materialistic wealth, people forgot about their health even when they are living in the age of AI. The world already faced corona pandemic and above all the poor life style raised the concern related to medical system and healthy life style. There are so many health issues, however, the patient and their life are profoundly impacted by diabetes mellitus, which has become one of the most common chronic diseases globally. Due to the limited resources, reliance on laboratory investigations, and clinical interpretation in traditional diagnostic processes, early-stage detection with prediction can be delayed in healthcare systems. The research paper gives a thorough framework for diabetes prediction based on machine learning. In this research work we use Using the Pima Indians Diabetes Dataset and made models using various supervised learning algorithms, such as Random Forest, SVM, Decision Tree, Logistic Regression, K-Nearest Neighbours, and XGBoost. We also employed Synthetic Minority Oversampling Technique (SMOTE)for minor data balancing and improved the results such as Logistic Regression is with 76.3%, Random Forest with 74.2%, Decision Tree with 68.8%, SVM with 64.2%, kNN with 70.4%, and Gradient Boosting with 77.5%.