Examining Different Machine Learning Approaches for Predicting Diabetes in the Early Stages

Many people around the world have diabetes, a common long-term health issue. It is characterized by elevated blood glucose levels and, if left untreated, may result in serious health problems. The accurate diagnosis and timely diabetes management are of utmost importance in preventing these complications. In this research, we utilize machine learning techniques to improve the diagnosis of diabetes. We employ eight different models, such as Logistic Regression, Bernoulli Naive Bayes, Support vector machine, Decision Trees, Multi-Layer Perceptrons, Random Forest, AdaBoost, XGBoost, and Stochastic Gradient Descent. The focus of this research is to implement hyperparameter tuning to optimize the performance of these models and rigorously evaluate their diagnostic accuracy using assessment criteria such as Accuracy, Recall, Precision, and lastly the F1-Score. Random Forest achieved the highest accuracy rate at 96.00% among the eight machine-learning models.