Prediksi Penyakit Stroke Menggunakan Metode Support Vector Machine

Abstract

Stroke is a neurological condition that results in impaired brain function and can cause disability and even death. In Indonesia, stroke is the leading cause of death after cancer and heart disease. Early detection and assessment of stroke plays an important role in reducing stroke mortality and severity. However, analyzing large amounts of medical record data is a challenge for doctors in diagnosing the disease effectively. In processing medical record data, machine learning can be used to find patterns and rules hidden in the data. One of the widely used machine learning methods is Support Vector Machine (SVM), which is well-known for its high accuracy in data classification. SVM has been used in various studies for clinical outcome prediction and medical diagnosis. This research uses the Support Vector Machine (SVM) method with optimized parameters and the application of the Synthetic Minority Over-sampling Technique (SMOTE) technique at the preprocessing stage to predict stroke disease. The Stroke Prediction Dataset is used to train and test the SVM model. Three different data splitting schemes are used, namely 70:30, 80:20, and 90:10, and k-fold cross validation is applied with k values of 5, 7, and 10 to comprehensively evaluate the model performance. The prediction results were evaluated with accuracy, precision, recall, and F1-score metrics. The highest accuracy was achieved using SVM with rbf kernel and 90:10 data split.