Automatic Recognition of Obstructive Sleep Apnoea Syndrome Using Power Spectral Analysis of Electrocardiogram and Hidden Markov Models

• DOI: 10.1109/issnip.2008.4762001
• Publication Date: 2009-01-28T11:25:17Z
• Authors: Tarik Al-ani; Chandan K. Karmakar; Ahsan H. Khandoker; Marimuthu Palanis...
• Keywords: Sleep

Obstructive sleep apnoea syndrome (OSA) is a very common disorder in breathing during sleep. OSA considered as clinically relevant when the breath stops more than 10 seconds and occurs five times per hour. In this work, we investigate noninvasive automatic approach to classify events based on power spectral analysis for feature extraction of ECG records hidden Markov models (HMMs). Based Bayesian inference criterion (BIC), proposed HMM training algorithm able select optimal number states corresponding each set features. For every state number, iteration initialized by most appropriate model using data clustering, rejection least probable previous iteration. Both off-line on-line schemes have been proposed. Only electrocardiogram (ECG) are detection OSA. preliminary report procedures validation results whole night digitized signals recorded from 70 subjects with normal obtained physionet database.