A New Algorithm to Score Apnea/Hypopnea Events based on Respiratory Effort Signal and Oximeter Sensors

• Published in: Journal of medical and biological engineering Vol. 40; no. 4; pp. 555 - 563
• Main Authors: Dell’Aquila, Carlos Rubén, Cañadas, Gabriel Eduardo, Laciar, Eric
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.08.2020; Springer Nature B.V
• Subjects: Air flow; Algorithms; Annotations; Apnea; Biomedical Engineering and Bioengineering; Cell Biology; Computer applications; Engineering; Imaging; Lie detectors; Original Article; Oxygen content; Radiology; Root-mean-square errors; Sensors; Statistical analysis; Thorax; I1 Health; Respiratory effort signals; SpO; Digital signals processing; I15 Health and economic development; SAHS
• ISSN: 1609-0985; 2199-4757
• DOI: 10.1007/s40846-020-00549-0

Purpose The aim of this work is to develop and evaluate a new algorithm for the detection of Apnea/Hypopnea events based on the AASM recommendations, with low computational cost and using alternative sensors, such as the respiratory effort in addition to the percentage of oxygen saturation (%SpO 2 ). Methods The algorithm proposed in this work uses as input signal the sum of the respiratory effort signals from thorax and abdomen, from which an estimate of the oronasal airflow signal is obtained. The latter is the primary sensor recommended by the AASM, being the effort signals considered as alternative sensors. The algorithm also uses the values of %SpO 2 in order to detect the events of Apnea/Hypopnea. In this work a method to estimate basal values of respiration and %SpO 2 is additionally proposed. The performance of the algorithm and both the alternative and recommended respiratory sensors are evaluated by two freely available PSG databases with annotations. Results For the algorithm, several statistical parameters are calculated (Sen%, Spe%, Acc%, Vpp%, Vpn%, coef. kappa) with whether the recommended (Resp FLOW ) (80.24; 98.00; 90.77; 96.49; 87.85; 0.80) or the alternative respiratory sensor (RIP FLOW ) (83.47; 94.76; 90.16; 91.61; 89.31; 0.79). The algorithm capacity to estimate the Apnea/Hypopnea Index (AHI) from Root Mean Square error (RMSe) is also evaluated, resulting RIPFlow (RMSe = 3.40) better than RespFlow (RMSe = 6.28). Conclusions The results obtained for both sensors were similar and comparable to other works using other sensors and more sophisticated processing techniques. In conclusion, the new algorithm proposed, together with the respiratory effort sensors and the %SpO 2 is adequate to be implemented in portable respiratory polygraphs.