results_summary_singlechannel.m

close all
clc
clear
addpath(genpath(pwd))

mydir = pwd;
idcs = strfind(mydir,filesep);
% second parent folder contains the datasets


model_name_all = {'dgdss', 'tiny', 'seq', 'x_joint', 'x_jointfuse','seqfused'};% {'dgdss', 'tiny', 'seq', 'x_joint'};
channel_num = 2;
sleepedf_num = 20;


for k = 1:length(model_name_all)

    model_name = model_name_all{1,k};
    load(['output_',model_name,'.mat'])

    % single-channel metrics + original deep system with multi-channel signals
    for ch = 1:3

        y_test_org = [];
        y_true_org = [];
        for i = 1:CV_number

            response = true_label{ch,i};
            response = response(:);
            if sum(response==0)>0
                response = response+1;
            end
            this_fold_number = fold_number{1,i};
            test_set = this_fold_number==i;
            [~,yhat]= max( softmax( hingeloss_traintest{ch,i}(:,test_set)')');
            y_test_org = [y_test_org;yhat(:)];
            y_true_org = [y_true_org;response(test_set)];
        end

        [acc, kappa , f1, sens, spec] = calculate_overall_metrics(y_true_org, y_test_org);
        acc_single(k,ch) = acc;
        f1_single(k,ch) = f1;
        kappa_single(k,ch) = kappa;

    end

    num_test(k,1)=length(response);
   
end

acc_single