This project aims to use machine learning models for diabetes early detection, a worldwide health concern. The dataset that was used comes from Kaggle and includes patient data, medical history, and diabetes status. To guarantee data quality, strict data preprocessing—which includes managing categorical variables—was carried out. Three machine learning models, implemented in Google Collaboratory, are compared in this study: Artificial neural networks, support vector machines, and K-nearest neighbors were used in the study. The problem of class imbalance was addressed with the synthetic minority over-sampling technique. To provide a thorough evaluation of each model's prediction ability for diabetes diagnosis, we included confusion matrices and other evaluation measures. Although the specific results are not provided here, they offer valuable insights into the relative performance of KNN, SVM, and ANN. This information can aid in selecting the most appropriate algorithm based on specific performance criteria relevant to the dataset in question.