Automatic assessment of an "apnoea-severity-factor" combined with heart rate analysis during polysomnographic examinations in infants

Infant apnoea classification in terms of severity and character is done subjectively by physicians based on polysomnographic recordings. Beside changes in respiratory parameters often significant drops in heart rate and O₂-saturation are observed during apnoea. The authors propose a new "apnoea-severity-factor", which is computed using apnoea duration patterns O₂-saturation values and heart rate variations during an apnoea. The beat-to-beat heart rate is calculated via interpolation techniques from a one channel ECG. Central, obstructive or mixed apnoea are identified and analyzed using a neural network. For each recorded apnoea its severity-factor is calculated automatically from the acquired signals and altogether statistically analyzed. First studies revealed, that the subjective assessment of medical experts showed good correlation with the authors\´ severity factor and apnoea duration parameters as well as O₂-values. An unexpected low correlation of heart rate variations against assessed values was found. Further investigations yielded an overall correlation coefficient of 0.9 between assessed and computed apnoea-severity-factors. It is concluded that an apnoea-severity-factor automatically computed from polysomnographic recordings and including the beat-to-beat heart rate variation can improve the subjective judgement of medical experts involved in SIDS prevention.