Morphological Prediction of CPAP Associated Acute Respiratory Instability (Self Similarity)

Abstract

This research presents a new algorithm to assess self-similarity (SS) as a signature of increased loop gain using respiratory effort signals, and demonstrates its utility in predicting acute failure of continuous positive airway pressure (CPAP) therapy. By analyzing effort signals from 2,145 split-night polysomnography studies, our model achieved strong predictive performance with area under the curve values of 0.82 and 0.84 for ROC and precision-recall curves, respectively. The SS metric combined with central apnea index and hypoxic burden outperformed conventional metrics alone, providing a more accurate, noninvasive approach to phenotyping obstructive sleep apnea for precision treatment strategies.

This data publication includes data and code needed to reproduce the results in this publication:

Nassi TE, Oppersma E, Labarca G, Donker DW, Westover MB, Thomas RJ. Morphological Prediction of CPAP Associated Acute Respiratory Instability. Ann Am Thorac Soc. 2024 Sep 17;22(1):138–49. doi: 10.1513/AnnalsATS.202311-979OC. Epub ahead of print. PMID: 39288402; PMCID: PMC11708763.

Background

Obstructive sleep apnea (OSA) is a condition associated with increased cardiometabolic and neurodegenerative disease risk. While traditionally viewed as primarily anatomical in nature, OSA involves multiple mechanisms including high loop gain (HLG), a key target for precision treatment. Existing loop gain measurement techniques are limited in clinical applicability. Our research addresses this gap by developing a self-similarity (SS) algorithm that can detect expressed HLG from standard respiratory signals, enabling better prediction of CPAP therapy outcomes and identification of patients who may benefit from adjunctive therapies targeting HLG.

Methods

We selected polysomnography recordings from the Massachusetts General Hospital Sleep Laboratory (2008-2022) archive. The MGH Institutional Review Board approved this retrospective analysis of clinically acquired PSG data (IRB #2014P000218). Abdominal respiratory inductance plethysmography signals were processed to detect self-similarity patterns in respiratory effort oscillations. A logistic regression model was trained using fivefold cross-validation to predict CPAP therapy effectiveness based on SS scores, with additional validation against an external dataset of home sleep apnea tests comparing SS with loop gain estimates.

Data Description

The dataset includes 2,145 unique split-night polysomnography recordings from the Massachusetts General Hospital, comprising respiratory effort signals, sleep staging, and scored respiratory events. All recordings included abdominal respiratory inductance plethysmography signals sampled at either 200 Hz or 512 Hz. Patient demographics include age, sex, race, and medication information. External validation used 202 home sleep apnea test recordings with both SS and phenotyping using PSG (PUP) loop gain estimates. Data provided here includes the intermediate files (extracted from raw edf PSG files), example .edf files. Code provided allows reproduction of the paper results and calculation of "intermediate files" from the raw PSG. 

Usage Notes

Data and code to generate all results and figures from the publication are provided here.
 

Ethics

This study of human subjects was approved by the Massachusetts General Hospital Institutional Review Board (IRB approval #2014P000218). The IRB provided a waiver of written consent for this retrospective analysis of clinically acquired data. All data has been deidentified.

Acknowledgements

This research was supported by U.S. National Institutes of Health grants RF1NS120947, R01AG073410, R01HL161253, R01NS126282, R01AG073598, R01NS131347, and R01NS130119 and National Science Foundation grant 2014431.

Conflicts of Interest

M.B.W. is a co-founder, scientific advisor, and consultant to Beacon Biosignals and has a personal equity interest in the company. Beacon Biosignals did not contribute funding and played no role in the study.

References

1. Nassi TE, Oppersma E, Labarca G, Donker DW, Westover MB, Thomas RJ. Morphological Prediction of CPAP Associated Acute Respiratory Instability. Ann Am Thorac Soc. 2024 Sep 17;22(1):138–49. doi: 10.1513/AnnalsATS.202311-979OC. Epub ahead of print. PMID: 39288402; PMCID: PMC11708763.