A Simple Risk Model of the Incidence of Atrial Fibrillation and Coronary Artery Disease ‎Using a Data Mining Algorithm: Risk Factor Prediction

Cardiovascular disease is one of the most important causes of death worldwide. Using data mining methods to create predictive models to identify people at risk to prevent complications from cardiovascular diseases will be very effective. The aim of this research is to predict the probability of infection in people with coronary heart disease and atrial fibrillation using support vector machine, neural network and decision tree algorithms based on factors affecting the disease.

This analytical research includes 300 records. The information required for this study was collected in 1400 using the records of patients admitted to Chamran and Khurshid hospitals in Isfahan. For data analysis, the information includes laboratory, demographic and family history sections using the CRISP method, the Cross Industry Standard Process for Data Mining (CRISP). Decision trees, neural networks and support vector machines are also used in the modeling section.

The sensitivity and specificity in the neural network data mining algorithm  are 87.5 and 71.11 respectively, 92.85 and 80 in the decision tree algorithm and 88.88 and 75 in the support vector machine. Therefore, the decision tree algorithm has a better performance for predicting the probability of heart and coronary artery diseases and atrial fibrillation. Also it was found that stress, high BMI, high blood pressure and type of job had the greatest effect on the occurrence of heart and coronary artery diseases and cardiac arrhythmias.

In the current study, the decision tree has the highest performance, so it can be used to determine the probability of coronary heart and vascular problems and atrial fibrillation.