Validation of a proprietary program and the Checkme™ O2 device for diagnosing obstructive sleep apnea in adults

The objective of this study was to evaluate the diagnostic performance of a proprietary program and the Checkme™ O2 device (ChO2) for diagnosing obstructive sleep apnea (OSA). For such purpose, we pre-selected 836 consecutive adult patients who underwent polysomnography in the sleep laboratory while simultaneously wearing ChO2. ChO2 data were analyzed with software that calculated various SpO2 variability parameters. The database was randomly split into an experimental group (75%, G-Exp) and a validation group (25%, G-Val). G-Exp was used to develop a binary logistic classification model (AHI < 5 or ≥ 5 events/h) named IMP-SpO2-N. Its diagnostic performance was assessed through K-fold cross-validation. A total of 566 patients were included (63% male; median age 52.5 years, median BMI 28.4 kg/m²). OSA prevalence (AHI ≥ 5 events/h) was 79.3%, with 58.6% having AHI ≥ 15 events/h. In G-Val, the IMP-SpO2-N achieved a mean sensitivity, specificity, positive/negative likelihood ratios, F1 score and accuracy of 92.50%, 92.3%, 12, 0.082, 95% and 92.4%, respectively. We conclude that our algorithm, based on dedicated software and measurements obtained from ChO2, showed high diagnostic accuracy for identifying adults with OSA.