Globally, heart disease is a leading cause of illness and mortality. This impacts people from all around the world. Accurate prediction of the risk of heart disease is crucial for early detection and prevention.For this, large amounts of features/attributes need to be stored and analyzed to diagnose a patient. Storing many features can lead to substandard management of data. We need to store only the chief features. In this study, we proposed a modified grey wolf optimizerfor feature selection.The resultant subset of features is then used to predict the risk of having a heart disease using machine learning model, Support Vector Machine (SVM). We compared the proposed algorithm with the existing GWO-SVM algorithm. We evaluated the effectiveness of the proposed algorithm using accuracy, sensitivity, and specificity metrics. Our results show that, using the modified grey wolf algorithm for feature selection and using SVM weobtained an accuracy of 95.82%, specificity of 94.64%, and sensitivity of 96.86%. The results show the proposed algorithm's capability for predicting the risk of heart disease and could contribute to the development of more accurate and efficient predictive models for heart disease risk assessment.