Accurate prediction models are needed to prevent and treat cardiovascular diseases (CVDs), a global health threat. This study recommends using SVM and ETLBO to improve CVD prediction accuracy. Age, gender, cholesterol levels, blood pressure, type of chest pain, electrocardiogram (ECG) results, and other risk variables are collected to start the study. Feature selection methods discover the most valuable predictors. SVM is the base classifier because it can handle complex and non-linear data connections. SVM requires manual hyperparameter adjustment, which is time-consuming and inefficient. The ETLBO technique automatically optimizes SVM hyperparameters to increase performance. The hybrid strategy is tested on a large CVD dataset against classic SVM and other optimization methods. The hybrid strategy surpasses SVM and other optimization strategies in accuracy, sensitivity, and specificity. In conclusion, the hybrid SVM-ETLBO methodology for CVD prediction outperforms classic SVM and other optimization methods.