Diabetes Diagnosis Using Machine Learning: A SVM-Based Approach

This article explores the use of Support Vector Machines (SVM) for diagnosing diabetes based on fourteen medical and behavioral variables. Following a theoretical overview of diabetes and SVM, a Python implementation is presented, including visualization of the hyperplane and margins through dimensionality reduction (PCA). The model stands out by training on the full set of variables without prior feature selection, ensuring complete exploitation of the available information. A practical case involving the insertion of a new patient is also addressed, illustrating the real-world application of the model. The achieved performance (accuracy, precision, recall) is evaluated and compared to that of other machine learning approaches, such as neural networks using the same dataset. The study concludes with a discussion on the results and perspectives for computer-assisted medical diagnosis.