SCREENING OF SLEEP APNEAS AND HYPOPNEAS THROUGH THE AUTOMATIC ANALYSIS OF MIDSAGITTAL JAW MOTION

This paper proposes a novel method for the scoring
of sleep apneas and hypopneas (SAHs) based on
the recording and the analysis of the midsagittal jaw
movements. Continuous wavelet transform was used
to delineate events which were likely to contain the
SAHs, while hidden markov models (HMMs)
classified events. Considering 28 recordings from
which awakenings were discarded, the method
detected SAHs with a sensitivity and a specificity of
82.2% and 78.3% respectively. Obstructive, central
and mixed respiratory events were distinguished
fairly accurately. The jaw motion is hence a reliable
marker of respiratory efforts and may suffice by itself
to score SAHs.