Abstract
The diagnosis of Obstructive Sleep Apnea (OSA) in children presents a challenging diagnostic problem given the high prevalence (2-3%), the resource intensity of the overnight polysomnography investigation, and the realisation that OSA poses a serious threat to the healthy growth and development of children. Previous attempts to develop OSA diagnostic systems using home pulse oximetry studies have failed to meet the accuracy requirements - particularly the low false normal rate (FNR) - required for a pre-PSG screening test. Thus the aim of this study is to investigate the feasibility of an OSA severity diagnostic system based on both oximetry and dual respiratory inductance plethysmography (RIP) bands. A total of 90 PSG studies (30 each of normal, mild/moderate and severe OSA) were retrospectively analyzed. Quantifications of oxygen desaturations (S), respiratory events (E) and heart rate arousals (A) were calculated and extracted and an empirical rule-based SEA classifier model for normal, mild/moderate and severe OSA defined and developed. In addition, an automated classifier using a decision tree algorithm was trained and tested using a 10-fold cross-validation. The empirical classification system showed a correct classification rate (CCR) of 0.83 (Cohen's Kappa κ=0.81, FNR=0.08), and the decision tree classifier achieved a CCR of 0.79 (κ=0.73, FNR=0.08) when compared to gold standard PSG assessment. The relatively high CCR, and low FNR indicate that a OSA severity system based on dual RIP and oximetry is feasible for application as a pre-PSG screening tool.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have