Download citation:
BibTeX | RIS | EndNote | Medlars | ProCite | Reference Manager | RefWorks
Send citation to:

---

Background: Discriminant analysis, as one of the classification methods, is one of the most practical statistical methods used in medical studies. In the case of classic statistical models which restricted to application, models such as artificial neural networks (ANNs) can be used for prediction and classification. In this study we compare the accuracy of ANN models against discriminant analysis and logistic regression models in Diagnosis of myocardial infarction.Methods: In this study the participants are 1000 case-control data, who suffered from Myocardial Infarction. Logistic regression, discriminant analysis and ANN models were fitted to the data. In ANN model three different algorithms used for training. The accuracy of models was compared using ROC analysis. SPSS, STATISTICA and SAS used for analysis.Findings: For quadratic discriminant method, prediction error percent, prediction correct percent, sensitivity, specificity and area under the ROC curve were 10.15, 89.85, 0.88, 0.90 and 0.92, respectively. Based on logistic regression method these measurements were 10.88, 89.12, 0.87, 0.91, and 0.94 , respectively. The results of ANN model showed that, these measurements were 3.97, 96.03, 0.95, 0.96 and 0.96, respectively. Between three training algorithms in ANN model, BFGS had the best performance.Conclusion: Findings demonestrad that the artificial nervation is more accurate for diagnosising Myocardial Infarction compared with logistic regression and quadratic discriminant methods. Key Words: Quadratic Discriminant, Logistic Regression, Artificial Nneural Network, Myocardial Infarction

Full-Text [PDF 512 kb]   (402 Downloads)    

Add your comments about this article : Your username or Email:

---