% run_classifier: This function takes a single record from the
% challenge training set, and uses that record together with the
% accompanying arousal annotations to tune model parameters
% appropriately. You may want to use this function as a basis for your
% own training code.
%
% Written by Mohammad Ghassemi and Benjamin Moody, 2018
function train_classifier(header_file_name)
% Read record info from the header file
data = parse_header(header_file_name);
X_tr = []; Y_tr = [];
%load all the the data associated with this subject
signals = load(data.signal_location); signals = signals.val;
arousal = load(data.arousal_location); arousal = arousal.data.arousals;
fs = str2num(data.fs);
n_samples = str2num(data.n_samples);
sid = data.subject_id;
signal_names = data.signal_names;
% select the window size and step size we want to use to
% compute features
window_size = 300 * fs;
window_step = 300 * fs;
% find the index of the SaO2 signal.
sao2_ind = find(contains(signal_names,'SaO2'));
% For each 'window', extract the variance of the SaO2
ind = 1;
for j = 1:window_step:n_samples-window_step
X_tr(ind) = var(signals(sao2_ind,j:j+window_step));
Y_tr(ind) = max(arousal(j:j+window_step));
ind = ind + 1;
end
% Set the -1 regions as 1 (treat unscored regions the same as
% arousals... you may not want to do this.)
toss = find(Y_tr == -1);
Y_tr(toss) = 1;
% Fit a logistic regression for each subject and save their model
display('Training Model...')
coeff = glmfit(X_tr,Y_tr','binomial');
% save the model for submission to challenge.i
display('Saving Model...')
save([sid '_model'],'coeff');