A K-Nearest Neighbour Classifier for Predicting Catheter Ablation Responses Using Noncontact Electrograms During Persistent Atrial Fibrillation

The mechanisms for the initiation and maintenance of atrial fibrillation (AF) are still poorly understood. Identification of atrial sites which are effective ablation targets remains challenging. Supervised machine learning has emerged as an effective tool for handling classification problems with multiple features. The main goal of this work is to use learning algorithms in predicting the responses of ablating electrograms and their effect on terminating AF and the cycle length changes. A total of 3,206 electrograms (EGMs) from ten persistent AF (persAF) patients were used. 5-fold cross-validation was applied, in which 80 % of the data were used as training set and 20 % used as validation. Dominant frequency (DF) and organisation index (OI) were calculated from EGMs (264 seconds) for all patients and used as input features. A k-nearest neighbour (KNN) classifier was trained using ablation lesion data and deployed in additional 17,274 EGMs that were not ablated. The classification accuracy of 85.2 % was achieved for the KNN classifier. We have proposed a supervised learning algorithm using DF features, which has shown the ability of accurately performing EGM signal classification that could be potentially used to identify ablation targets and become a robust real-time patient diagnosis system.