New feature set for better representation of dynamic of RR intervals in Poincare plot

Traditional features extracted from Poincare plot (e.g., SD1 and SD2) ignore temporal information and only quantify point distribution. In this article, a new feature set is proposed to capture the dynamic of point distribution in the Poincare plot. For quantification of temporal information in Poincare plot, extracted ADP features (angle, direction, and position) from point distribution in the Poincare plot were used with K-Nearest Neighbor (KNN) classifier for classification of four cardiac condition obtained from Physionet database (normal sinus rhythm, myocardial infarction, congestive heart failure, and atrial fibrillation). KNN was trained on 70% of data as a train set, and the accuracy was evaluated on 30% of data as a test set. Accuracy of classification were 94.8% and 95.58% for training and test set, respectively. Furthermore, ADP features were used for creating a new map that represents the temporal information of points in Poincare plot.