A Comparison between the SVM and ANN Techniques for Predicting Coronary Artery Diseases
Publish place: The 2nd Medical Informatics Conference and the 7th Electronic Health Conference and ICT Applications in Iranian Medicine
Publish Year: 1397
نوع سند: مقاله کنفرانسی
زبان: English
View: 412
نسخه کامل این Paper ارائه نشده است و در دسترس نمی باشد
- Certificate
- من نویسنده این مقاله هستم
این Paper در بخشهای موضوعی زیر دسته بندی شده است:
استخراج به نرم افزارهای پژوهشی:
شناسه ملی سند علمی:
NCMIMED02_005
تاریخ نمایه سازی: 1 دی 1397
Abstract:
Background: Currently, cardiovascular diseases (CVDs) are the first leading cause of death worldwide. World health organization has estimated that due to CVDs, the morality rate will mount to 23 million cases by 2030. Similarly, in Iran, a considerable growth of heart disease has been reported by the Ministry of Health. Hence, data mining can help improving the precision and accuracy of Coronary Artery Disease (CAD) predictions. The objective of the study was to compare the accuracy of the CAD predictions made by ANN and SVM techniques. Material and Methods: The present study was conducted via descriptive-analytical method. The study population included all the CAD patients hospitalized in three hospitals affiliated to AJA University of Medical Sciences betweenMarch 2016 and March 2017. Totally, 1324 records with 26 characteristics affecting the CAD incidence were extracted from hospital databases. Initially, normalizing, processing, and cleaning of the data were finished and a database specifically designed for the data was created in SPSS V23.0 & Microsoft Excel 2013 was used to format data to the R3.3.2 data mining software. Subsequently, the characteristics affecting the prediction of the variables were extracted. Results: Based on the findings, the most important variables affecting the CAD incidence included gender, age, weight, job, living place, family history, smoking, associated disease, average heart rate, triglyceride & creatinine level, and chest pain. The results obtained from data mining algorithms indicated that the SVM with lower MAPE (112.03) and higher Hosmer-Lemeshow statistic (16.71) yielded better fitness of data. Furthermore, considering the higher level of sensitivity (92.23) and characteristics (74.42), it was found out that the SVM predicted the CAD with higher strength and sensitivity than ANN. Finally, since the area under the ROC curve in SVM was more than that in ANN, it could be concluded that this model had higher accuracy compared to the ANN model. Conclusion: According to results, the SVM featured higher diagnostic accuracy and exhibited better performance than the ANN. Furthermore, the SVM was characterized with higher accuracy and it provided a better classification for prediction of CAD.
Keywords:
Coronary Artery Disease (CAD) , Data mining , Artificial Neural Network (ANN) , Support Vector Machine (SVM)
Authors
Haleh Ayatollahi
Associate Professor, Department of Health Information Management, School of Health Management and Information Sciences, Iran University of Medical Sciences, Tehran, Iran
Leila Gholamhosseini
PhD Student, School of Health Management and Information Sciences, Iran University of Medical Sciences, Tehran, Iran
Masoud Salehi
Associate Professor, Department of Biostatistics, School of Public Health, Iran University of Medical Sciences, Tehran,