Enhancing Early Detection of Diabetes Using Machine Learning Algorithms

Diabetes mellitus, a chronic metabolic disorder, presents a significant global health problem with increasing prevalence. Early detection of disease is crucial for effective management and the prevention of complications. This research investigates the application of machine learning (ML) algorithms in diabetes prediction, to assess their effectiveness through analysis of a diverse dataset of 2000 individuals. Numerous ML techniques are used in this study, such as Naive Bayes, Logistic Regression, Support Vector Machine, Random Forest, K-Nearest Neighbors, and Gradient Boosting. The dataset goes through appropriate preprocessing, data classification, and model building. Evaluation metrics include accuracy, receiver operating characteristic - area under the curve (ROC AUC), and interpretability of the models. The results demonstrate varied performance across algorithms. Random Forest emerged as the most successful, achieving 94.2% accuracy and a robust ROC-AUC of 92.1%. Logistic Regression and SVM exhibit moderate performance, while KNN and Gradient Boosting show notable improvements. The study contributes to understanding ML’s role in diabetes prediction and emphasizes the importance of model selection in healthcare analytics.