The Comparative Early Prediction Model for Cardiovascular Disease Using Machine Learning

Abstract

Cardiovascular disease (CVD) is a leading cause of death and a major contributor to disability. Early detection of cardiovascular disease using ANFIS has the potential to reduce costs and simplify treatment. This study aims to develop a prediction model using ANFIS (Adaptive Neuro-Fuzzy Inference System) for early detection of cardiovascular disease. The dataset used consists of 500 data with 12 features, including various risk factors such as blood sugar levels, cholesterol, uric acid, systolic blood pressure, diastolic blood pressure, body mass index (BMI), age, smoking habits, lifestyle, genetic factors, and gender, and one label feature. This study compares cardiovascular disease prediction models using machine learning methods, namely Support Vector Machine (SVM), K-Nearest Neighbor (K-NN), and ANFIS. The development of the KNN algorithm involves the value of K=5 with the Euclidian distance measure. The SVM algorithm used a kernel cache of 200 and a convergence epsilon of 0.001. The ANFIS model was built using 500 data sets divided into training (70%) and testing (30%) data, with learning rate variations of 0.01, 0.05, 0.1, 0.2, and 0.5. The results of testing the early detection model show for SVM, the accuracy value is 0.760, the precision value is 0.839, and the recall value is 0.671. For the KNN model, the accuracy value is 0.758, the precision value is 0.768, and the recall value is 0.771. As for the ANFIS model, the accuracy value reaches 0.989, precision value 0.996, and recall value 0.988. The model using ANFIS has the highest performance. Further study of the model using ANFIS with learning rate variations shows that a learning rate of 0.1 provides the most optimal performance.