Today, the rapid development of artificial intelligence has increased the frequency of use of this technology in many sectors. With this technology, which is in great demand in the health sector, it is used in many different areas such as diagnosis of diseases, appropriate treatment methods, nutrition plan according to the patient. Cardiovascular diseases are complex diseases that seriously affect human life. In the study using the publicly available Framingham dataset, there are 4240 observations and 16 variables. The number of variables was increased to 32 by feature extraction on the dataset. While creating the variables, some features in the Framingham risk map were taken into account. In addition, the unbalanced distribution in the dependent variable was eliminated. After these procedures, the risk of heart attack was predicted with LR, SVM, RF, AdaBoost, and Extra Tree algorithms from machine learning models. When the results of the training set and test set are analyzed, the best results are given by Random Forest with 94% and Extra Tree algorithms with 93%. The best increase was obtained in SVM with logistic regression.