Estimation of Risk Factors Related to Heart Attack with Xgboost That Machine Learning Model

Abstract

Objective: The objective of this work is to classify heart attack cases using the open-access heart attack dataset and one of the machine learning techniques called XGBoost. Another aim is to reveal the risk factors associated with having a heart attack as a result of the modeling and to associate these factors with heart attack.Methods: In the study, modeling was done with the XGBoost method using an open access data set including the factors associated with heart attack. Model results were evaluated with accuracy, balanced accuracy, specificity, positive predictive value, negative predictive value, and F1-score performance metrics. In addition, 10-fold cross-validation method was used in the modeling phase. Finally, variable importance values were obtained by modeling. Results: Accuracy, balanced accuracy, sensitivity, specificity, positive predictive value, negative predictive value, and F1 score from by XGBoost modeling were 89.4%, 89.4%, 88.4%, 90.3%, 88.4%, 90.3%, and 88.4%, respectively. According to the variable importance values obtained for the input variables in the data set examined in this study, thal2, oldpeak, thal3, ca1, and exang1 were obtained as the most important variables associated with heart attack.Conclusions: With the machine learning model used, the heart attack dataset was classified quite successfully, and the associated risk factors were revealed. Machine learning models can be used as clinical decision support systems for early diagnosis and treatment.