AUTHOR=Maresh Scott , Athikumar Adhithi Keerthana , Ahmed Nabila , Chandu Shivapriya , Prowting Joel L. , Tumah Layth , Najjar Abed A. , Khan Hamza , Sankari Muna , Lasisi Oluwatobi , Ravelo Laurel A. , Peppard Paul E. , Badr M. Safwan , Sankari Abdulghani TITLE=Role of automated detection of respiratory related heart rate changes in the diagnosis of sleep disordered breathing JOURNAL=Frontiers in Sleep VOLUME=2 YEAR=2023 URL=https://www.frontiersin.org/journals/sleep/articles/10.3389/frsle.2023.1162652 DOI=10.3389/frsle.2023.1162652 ISSN=2813-2890 ABSTRACT=Study objectives

The objective of this study was to determine whether electrocardiogram (ECG) and heart rate accelerations that occur in the vicinity of respiratory events could predict the severity of sleep-disordered breathing (SDB).

Methods

De-identified polysomnogram (NPSG) recordings from 2091 eligible participants in the Sleep Heart Health Study (SHHS) were evaluated after developing and validating an automated algorithm using an initial set of recordings from 1,438 participants to detect RR interval (RRI) dips in ECG and heart rate accelerations from pulse rate signal. Within-subject comparisons were made between the apnea-hypopnea index (AHI) and both the total RRI dip index (total RRDI) and total heart rate acceleration index (total HRAI).

Results

The estimated AHIs using respiratory-related HRAI correlated with NPSG AHI both in the unadjusted and adjusted model (B: 0.83 and 0.81, respectively P < 0.05). Respiratory-related HRAI had a strong agreement with NPSG AHI (intraclass correlation coefficient-ICC: 0.64, whereas respiratory-related RRDI displayed weaker agreement and ICC: 0.38). Further assessment of respiratory-related HRAI (≥5 events/h) showed a strong diagnostic ability (78, 87, 81, and 56% agreement for traditional AHI cutoffs 5, 10, 15, and 30 events/h, respectively). At the AHI cutoff of 5 events/h the receiver operating curves (ROC) revealed an area under the curve (AUCs) of 0.90 and 0.96 for RE RRDI and RE HRAI respectively.

Conclusion

The automated respiratory-related heart rate measurements derived from pulse rate provide an accurate method to detect the presence of SDB. Therefore, the ability of mathematical models to accurately detect respiratory-related heart rate changes from pulse rate may enable an additional method to diagnose SDB.