Implementasi Algoritma Random Forest, Logistic Regression, dan Gradient Boosting untuk Deteksi Dini Diabetes Mellitus

Abstract

Diabetes Mellitus is a chronic metabolic disease with an increasing prevalence worldwide, including in Indonesia. Early detection is important to reduce the risk of complications and support preventive health actions. This study aims to develop a web-based early diabetes risk screening application using machine learning algorithms, namely Random Forest, Logistic Regression, and Gradient Boosting. The dataset used in this study is the CDC Diabetes Health Indicators dataset, which contains demographic, health condition, and lifestyle variables related to diabetes risk. The research process includes data collection, preprocessing, model training, model evaluation, and implementation into a web-based application. Model performance was evaluated using accuracy, precision, recall, and F1-score. The results show that Gradient Boosting achieved the best performance with an accuracy of 84.52%, followed by Random Forest with 83.20% and Logistic Regression with 81.88%. The developed application is able to provide diabetes risk scores, risk categories, health recommendations, AI-based consultation, and screening history. This study shows that machine learning can be used as a supporting tool for early diabetes risk screening.