Detection of Respiratory Events by Respiratory Effort and Oxygen Desaturation

Sleep respiratory events are scored based on the reduction of airflow measured by a thermistor or nasal pressure cannula, together with oxygen desaturation and arousal criteria for hypopneas. We investigated whether automatic scoring can be performed without using the uncomfortable oronasal sensors and developed an automatic scoring system that is compatible with level III home sleep apnea testing devices. We developed a respiratory event detection algorithm, based on SpO2 and respiratory effort signal measured from respiratory inductance plethysmograph (uncalibrated RIPsum), that outputs the time and duration of detected events and calculates an apnea–hypopnea-index (AHI) based on total recording time. The algorithm was tested on 98 polysomnography (PSG) recordings of patients, 77 with suspected sleep apnea and 21 without. The results were compared to annotations provided by the PSG systems where PSG AHI was computed using the total sleep time. The predicted AHI was evaluated for correlation and agreement with the PSG AHI using the intra-class correlation coefficient (ICC). Severity classification was performed and evaluated using the following categories: normal (< 5), mild (5–15), moderate (15–30), and severe ($$\ge 30$$). The ICC between predicted AHI and PSG AHI scored r = 0.96 (0.95–0.97, p < 0.001). The algorithm correctly predicted the severity for 74 recordings, overestimated 16, and underestimated 8. There was no misclassification by more than one severity level. Using respiratory effort and SpO2, our algorithm was able to detect respiratory events with high correlation and agreement compared to full PSG-based detection.