Novel Type IV Monitoring Device With Advanced Oximetry Indicators Offers Accurate Diagnosis of Obstructive Sleep Apnea in Adults

Abstract Background: As a Type IV sleep monitoring device, PM50-B is an integrated high-resolution oximeter with an accelerometer, which can discern the oxygen desaturation due to motion artifacts. The aims of this study were to develop an advanced algorithm based on pulse oximetry of Berry PM50-B and determine optimal oximetry indicator for detecting suspected obstructive sleep apnea (OSA) in adult patient. Methods This was an observational prospective study. Suspected OSA patients underwent the PM50-B device in parallel with standard diagnostic methods (Type I polysomnography [PSG], Type II PSG, or Type III home sleep apnea test [HSAT]). The apnea-hypopnea index (AHI) was analyzed with the criteria recommended by the American Academy of Sleep Medicine (AASM) for the definition and classification of the severity of OSA. Compatible with Python programming, the raw data from the PM50-B device generates a series of oximetry parameters. Such as oxygen desaturation index (ODI) with a ≥4% drop sustained for 8 seconds (ODI4_8), the cumulative percentage of the time spent at a SpO2 less than 90% (CT90), and average of the lowest 1% of nocturnal SpO2 values (LSpO2_Min1%Avg). The program conducts Bland-Altman plots, Pearson correlation, and receiver operating characteristics (ROC) for statistical evaluation. Furthermore, we established a diagnostic model to identify OSA using cut-offs of 5, 15, and 30 for ODIs. For other oximetry indicators the optimal cut-offs were determined using the Youden index. Results Of 499 adults enrolled in the study, the ODI4_8 had a better correlation with the AHI (root mean square error 6.342). The intraclass correlation coefficient for ODI4_8 vs AHI, ODI3_10 vs AHI was 0.956 and 0.951.With a threshold of ≥15, ODI4_8 showed 90.32% sensitivity, 93.26% specificity for detecting moderate-to-severe OSA. Diagnostic performances with the best cutoff for the evaluated pairs were compared: LSpO2_Min1%Avg vs LSpO2 and LSpO2_Min0.5%Avg, CT90 vs CT93 and CT87. For LSpO2_Min1%Avg with a cut-off of ≤ 85.56, it had 84.33% sensitivity, 86.88% specificity for detecting moderate-to-severe OSA, and its accuracy was superior to that of LSpO2 (78.76%) or LSpO2_Min0.5%Avg (83.97%). The AUC for the diagnosis of moderate-to-severe OSA determined by ODI4_8, LSpO2_Min1%Av and CT90 was 0.977, 0.955 and 0.964, respectively. Conclusion These oximetry indicators derived from PM50-B device exhibit good sensitivity and specificity. The small, comfort, and easy-to-use device based on advanced oximetry indicators is more accurate for OSA screening in a large population when standard diagnostic tests are unavailable or not tolerated.