KLASIFIKASI ISYARAT DETAK EKG PENYAKIT ARRHYTHMIA BERBASIS CASCADE TRANSPARENT CLASSIFIER

• المؤلفون الرئيسيون: , Anugerah Galang P, , Noor Akhmad Setiawan S.T., M.T., Ph.D
• التنسيق: Theses and Dissertations NonPeerReviewed
• منشور في: [Yogyakarta] : Universitas Gadjah Mada 2014
• الموضوعات: ETD
• الوصول للمادة أونلاين: https://repository.ugm.ac.id/133654/; http://etd.ugm.ac.id/index.php?mod=penelitian_detail&sub=PenelitianDetail&act=view&typ=html&buku_id=74401

Arrhythmia is an illness often encountered in patients with cardiac problems. Its symptoms can be observed using an electrocardiogram (ECG). Automatic detection system based on machine learning has been developed to help doctor discover arrhythmia symptoms. Unfortunately, rarely of them are capable of explain knowledge behind the decision being taken. Transparent classification system is needed to be developed. In order to increase human�s understanding of knowledge embedded in the system. To achieve these goals, a method was designed. Firstly, EGC signals were separated, then each signal was extracted using a feature extraction method. Furthermore, several of extracted feature�s attribute were selected, and the last step was classifying data using decision tree and the rule based classifier. Classification performance was improved by implementing a cascade structure, which used two or more classification methods to classify the data. Performance of classifications was tested using 10-fold cross validation, consider only by its average of both accuracy and the number of rule required. Best results, taking into consideration the accuracy and rules, obtained using rule-based classifier, with 92.4% average accuracy and required 40 rules. Best accuracy obtained by implementing a cascade classifier, with 92.841% accuracy and requires 80 rules. As a conclusion, transparent classifier is able to perform with reasonable accuracy compare with previous research, using only relatively small amount of rules.