Diabetes prediction model using machine learning techniques

Diabetes has emerged as a significant global health concern, contributing to various severe complications such as kidney disease, vision loss, and coronary issues. Leveraging machine learning algorithms in medical services has shown promise in accurate disease diagnosis and treatment, thereby alleviating the burden on healthcare professionals. The field of diabetes forecasting has rapidly evolved, offering the potential for early intervention and patient empowerment. To this end, our study presents an innovative diabetes prediction model employing a range of machine learning techniques, including Logistic Regression, SVM, Naïve Bayes, and Random Forest. In addition to these foundational techniques, we harness the power of ensemble learning to further enhance prediction accuracy and robustness. Specifically, we explore ensemble methods such as XGBoost, LightGBM, CatBoost, Adaboost, and Bagging. These techniques amalgamate predictions from multiple base learners, yielding a more precise and resilient final prediction. Our proposed framework is developed and trained using Python, utilizing a real-world dataset sourced from Kaggle. Our methodology is rigorously examined through performance evaluation metrics, including the confusion matrix, sensitivity, and accuracy measurements. Among the ensemble techniques tested, CatBoost emerges as the most effective, boasting an impressive accuracy rate of 95.4% compared to XGBoost's 94.3%. Furthermore, CatBoost's higher AUC-ROC score of 0.99 reinforces its potential superiority over XGBoost, which achieved an AUC-ROC score of 0.98.