Framework to Predict Diabetes Using Boruta and Genetic Algorithm as Feature Selector

Diabetes mellitus is one of the chronic diseases that poses a significant threat to human health. Therefore, timely prediction is crucial to mitigate its effects and enable prompt medical intervention. The objective of this study is to propose a hybrid predictive model that combines ensemble learning with feature optimization using Boruta and a Genetic Algorithm for feature selection. The Boruta utilizes the random forest algorithm to rank features and remove irrelevant ones. In addition, the Genetic Algorithm selects the optimal subset of features to improve model performance. The final classification employs ensemble models, specifically stacking, which leverages the strengths of different base classifiers. The model is evaluated on two well-known benchmark datasets: the PIMA Indians Diabetes Dataset and the Frankfurt Diabetes Dataset. Experimental results will be evaluated to assess whether the proposed model achieves superior performance in key metrics— accuracy, precision, recall, and AUC—compared to baseline classifiers, aiming to make it a reliable tool for early diabetes prediction. Disease, Diabetes, Feature Selection, Boruta Algorithm, Genetic Algorithm