TWO ALTERNATIVES FOR ATOMS SELECTION APPLIED TO SCREENING FOR SLEEP DISORDERS

ROMAN ROLON; LEANDRO DI PERSIA; HUGO RUFINER; RUBEN SPIES

Tandil

Congreso; V Congreso de Matemática Aplicada, Computacional e Industrial; 2015

The Obstructive Sleep Apnea-Hypopnea Syndrome is characterized by repetitive episodes of upper airway obstruction that occur while sleeping, usually associated with a reduction in blood oxygen saturation (SaO2). In this article, application of sparse representations of SaO2 signals over a subcomplete dictionary for classification tasks is discussed. A sparse representation describes an SaO2 signal in terms of a linear combination of a few columnsof a previously learned dictionary. The SaO2 signals are used in order to predict the occurrence of apnea-hypopnea events. A dictionary is learned by using a statistical method. Then a greedy pursuit algorithm is used in order to find the solution of a linear inverse problem with sparse constraint. The sparse vectors are used as input of a multilayerperceptron neural network. Finally an apnea-hypopnea index is estimated to grade the severity of OSAHS. Different alternatives for exploiting the activations of most discriminative atoms are evaluated.