| Author/Editor | Grošelj, Ciril | |
| Title | Sistem nevronskih mrež v diagnostiki ishemične bolezni srca | |
| Type | monografija | |
| Place | Ljubljana | |
| Publisher | Medicinska fakulteta | |
| Publication year | 1999 | |
| Volume | str. 98 | |
| Language | slo | |
| Abstract | Because of insufficient diagnostic abilities any rationalization of coronary artery disease (CAD) diagnostic process would be useful. Artificial intelligence method "Machine learning" proved to be useful for making decision in many fields of science and also in medicine. The diagnostic accuracy of such decisions usually rises. We proved in our study the applicability of the method in the diagnostic decisions for CAD. We tested a few reported Machine learning methods and according to preliminary results choose the method named Naive Bayesian classifier. With this method we did our final calculations. Our thesis that the diagnostic accuracy of stepwise diagnostic process for CAD will rise with application of such method we proved in a group of 327 patients, who were in diagnostic process for CAD. The results after each diagnostic step were expressed in term of accuracy. On each diagnostic step, we compared the accuracy get in usual way with that, by Naive Bayes classifier. We also tested the both approaches to diagnostic deciding, the standard and the probabilistic one bay comparing the results between each other and with machine learning results. According to our results, diagnostic deciding by Naive Bayesian classifier rises the diagnostic accuracy, so by standard as probabilistic approache. Ranging the diagnostic decision methods, is far on top classical approach implemented by Machine learning, followed by implemented probabilistic one, than standard approach and far behind the usual probabilisic. In conciusion the Machine learning method Naive Bayesian classifier proved to be a well learnable and good applicable method for diagnostic decisions making in diagnosing CAD. | |
| Descriptors | MYOCARDIAL ISCHEMIA ARTIFICIAL INTELLIGENCE BAYES THEOREM CORONARY ANGIOGRAPHY NEURAL NETWORKS (COMPUTER) SENSITIVITY AND SPECIFICITY EXERCISE TEST ELECTROCARDIOGRAPHY PHYSICAL EXAMINATION MEDICAL HISTORY TAKING |