Abstract:

This research investigates the use of machine learning algorithms for early detection of diabetes. Due to its global prevalence and significant impact on health, timely identification of diabetes is crucial for effective treatment. In this study, machine learning models including Gradient Boosting Machines, Extreme Gradient Boosting, Light gradient-boosting machine, Categorical Boosting, k-Nearest Neighbors, Random Forest, Ridge Classifier, Logistic Regression, Gaussian Naive Bayes, and Decision Tree are utilized to assess their capabilities in diabetes diagnosis. The primary aim is to train these models to distinguish between individuals with diabetes and those without, using relevant features from the dataset. Since the classes in the dataset are imbalanced, the SMOTE technique is applied to improve model performance. Categorical Boosting achieved the highest accuracy rate of 90.05%, making it the most successful model. By systematically evaluating the performance of these prominent machine learning models, valuable insights can be gathered regarding their ability to recognize complex patterns indicative of diabetes. As a result, healthcare professionals and researchers can leverage this newfound understanding to develop more accurate and effective diagnostic tools, enabling early intervention and subsequently improving the overall quality of life for individuals affected by diabetes.

Keywords: