HRV feature selection based on discriminant and redundancy analysis for neonatal seizure detection

This paper addresses the feature selection problem by using a discriminant and redundancy based method to select a feature subset with high discriminatory power between the classes of newborn heart rate variability (HRV) corresponding to seizure and non-seizure. The proposed method combines the Fast Correlation Based Filter (FCBF) criteria for redundancy analysis with the area under the Receiver Operating Curves (AUC) for discriminant analysis. The classification accuracies of the selected features were compared using 3 different classifiers, namely linear classifier, quadratic classifier and k-Nearest Neighbour (k-NN) statistical classifiers in a leave-one-out (LOO) cross validation. It was found that the 1-NN outperformed the other classifiers resulting in a significant reduction in feature dimensionality while achieving 85.7% sensitivity and 84.6% specificity.