Binary Autonomic State Classification based on Basic Heart Rate Variability Indices
This paper examined the basic state classification accuracy utilizing standard heart rate variability (HRV) indices. HF, LF/HF, HRV mean and variance were utilized for the binary classification of four states, i.e. Control, sympathetic and/or para-sympathetic activity blocked state. Support Vector Machine (SVM) and the three layer neural network (NN) achieved average pairwise correct classification accuracy of 0.931 and 0.886 respectively for the case of utilizing normalized indices. The result suggests that SVM may be a useful building block for more precise state classification introducing a binary tree of the autonomic states.