A Bayesian neural network approach for sleep apnea classification

In this paper a method for sleep apnea classification is proposed. The method is based on a feedforward neural network trained using a bayesian framework and a cross-entropy error function. The inputs of the neural network are the first level-5-detail coefficients obtained from a discrete wavelet transformation of the samples of the thoracic effort signal corresponding to the apnea. In order to train and validate the presented method, 120 events from 6 different patients were used. The true error rate was estimated using a 10-fold cross validation. The presented results were averaged over 100 different simulations and a multiple comparison procedure was used to model selection. The mean classification accuracy obtained over the test set was 83.78% +/- 1.90.