An Optimized Bagging Ensemble Learning of Machine Learning Algorithms for Early Detection of Diabetes

Diabetes is a common and serious medical illness that needs to be properly and promptly diagnosed in order to be managed. This study proposes an optimized Bagging Classifier, a novel method for predicting cases of diabetes and non-diabetes using machine learning methods. Based on a variety of evaluation metrics, the Logistic Regression, Support Vector Machine (SVM), and Random Forest techniques are compared to the optimized Bagging Classifier's performance. With an astounding accuracy of 93%, the suggested optimized Bagging Classifier achieves outstanding results. A significant percentage of real diabetes cases are correctly identified among the projected positives with an accuracy of 81%. In order to ensure prompt medical care, the perfect recall of 100% demonstrates its capacity to record all actual incidents of diabetes. The reliability in achieving a harmonious balance between recall and precision is further supported by the balanced F1 score of 0.89. Additionally, its better capacity to discriminate between cases with and without diabetes is shown by the high AUC score of 0.94. In terms of accuracy, precision, recall, F1 score, and AUC score, the optimized Bagging Classifier surpasses the other algorithms, according to the comparison analysis. Although the other models perform admirably, recall and F1 score are weak, which could result in missed instances of diabetes. This study emphasizes the importance of the optimized Bagging Classifier in accurately predicting diabetes and demonstrates its potential to assist medical professionals in making defensible decisions. The critical assessment emphasizes the need for more optimization and modification of the other algorithms to improve their prediction powers. Proposed research provides useful information about the efficacy of different machine learning algorithms for diabetes prediction, enabling better patient outcomes and healthcare administration.