Comparative Study of Diabetes Prediction using Machine Learning Approaches

Abstract

Background: Diabetes is a rapidly growing global health problem. Diabetes mellitus is a chronic disease that occurs when one’s pancreas no longer able to produce enough insulin. The long-term hyperglycemia during diabetes causes chronic damage and dysfunction of various tissues, especially the eyes, kidneys, heart, blood vessels, and nerves. The intention of using ML in healthcare is to increase the diagnostic accuracy and effectiveness of therapy and help clinicians in their practice of patient management with improved outcomes. Disease prediction using ML is gaining significant attention for healthcare

Methods: This study explores the use of six supervised machine learning algorithms to predict diabetes using the Pima Indians Diabetes Database (PIDD). Several well-known ML techniques were implemented and compared, including Support Vector Machine (SVM), Naïve Bayes (NB), K-Nearest Neighbor (KNN), Random Forest (RF), Logistic Regression (LR), and Decision Tree (DT). The performance of the models was assessed using prediction accuracy.

Results: The results of study indicate Support Vector Machine (SVM) outperformed the other models. SVM obtained a prediction accuracy of 74%, outperforming the other algorithms.

Conclusion: The findings suggest that machine learning methods can significantly improve early diabetes prediction. SVM provides best predictive ability for the given dataset. This study demonstrates that machine learning–based systems can assist healthcare professionals in making early diagnoses and informed decisions, thereby helping to prevent serious complications associated with diabetes.

Implication: This work can help healthcare facilities understand the usefulness and use of machine learning algorithms in early diabetes prediction.