EEG-Based Automatic Sleep Staging Using Ontology and Weighting Feature Analysis

Algorithm 1

Finding the best threshold value.

Input:pbest = 0, ACbest = 0, Sfemale = , Smale = ; % Initialize the optimal threshold value, the highest accuracy rate, the best subset of features.
(1)	for p = [0.0005: 0.0005:1]
(2)	S1, S2; % Two sets of feature subsets for different genders were produced based on correlation analysis.
(3)	S1, classifier; % S1 and S2 were forwarded to the RF classifier to classify five-state sleep stages.
(4)	AC; %The classification accuracy obtained by RF.
(5)	if ACbest ≤ AC
(6)	ACbest = AC, pbest = p, Sfemale= S1, Smale = S2;
(7)	end
(8)	end
Output:pbest, Sfemale, Smale;