Wavelet-based Gaussian-mixture hidden Markov model for the detection of multistage seizure dynamics: A proof-of-concept study

Table 3 Performance measure for supervised and unsupervised seizure detection approach

Using identical wavelet coefficient features, the performance of the unsupervised HMM seizure detector (HMM_opt7D) did not show any statistically significant improvement over the supervised approach (WANN). However, when the rate of change of wavelet information was included in the feature space, the optimality index of HMM_opt14D was significantly better than both WANN and HMM_opt7D.

ISSN: 1475-925X