Comparative analysis of predictive methods for early assessment of compliance with continuous positive airway pressure therapy

Abstract

Background: Patients suffering obstructive sleep apnea are mainly treated with continuous positive airway pressure (CPAP). Although it is a highly effective treatment, compliance with this therapy is problematic to achieve with serious consequences for the patients' health. Unfortunately, there is a clear lack of clinical analytical tools to support the early prediction of compliant patients.

Methods: This work intends to take a further step in this direction by building compliance classifiers with CPAP therapy at three different moments of the patient follow-up, before the therapy starts (baseline) and at months 1 and 3 after the baseline.

Results: Results of the clinical trial shows that month 3 was the time-point with the most accurate classifier reaching an f1-score of 87% and 84% in cross-validation and test. At month 1, performances were almost as high as in month 3 with 82% and 84% of f1-score. At baseline, where no information of patients' CPAP use was given yet, the best classifier achieved 73% and 76% of f1-score in cross-validation and test set respectively. Subsequent analyzes carried out with the best classifiers of each time point revealed baseline factors (i.e. headaches, psychological symptoms, arterial hypertension and EuroQol visual analog scale) closely related to the prediction of compliance independently of the time-point. In addition, among the variables taken only during the follow-up of the patients, Epworth and the average nighttime hours were the most important to predict compliance with CPAP.

Conclusions: Best classifiers reported high performances after one month of treatment, being the third month when significant differences were achieved with respect to the baseline. Four baseline variables were reported relevant for the prediction of compliance with CPAP at each time-point. Two characteristics more were also highlighted for the prediction of compliance at months 1 and 3.

Trial registration: ClinicalTrials.gov Identifier, NCT03116958 . Retrospectively registered on 17 April 2017.

Keywords: Continuous positive airway pressure; Machine learning; Obtrusive sleep apnea; Predictive methods.

Conflict of interest statement

Ethics approval and consent to participate

Ethics approval was obtained from the Human Research Ethics Committee of University Hospital of Arnau de Vilanova (Lleida) with the reference number 10/2014, in accordance with the Declaration of Helsinki, and written informed consent to participate was obtained from each patient included in the investigation.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Figures

Fig. 1

Pipeline steps designed for building classifiers for compliance with the CPAP therapy

Fig. 2

ROC curves for cross-validation and test of the best pipeline for dataset D0

Fig. 3

ROC curves for cross-validation and test of the best pipeline for dataset D1

Fig. 4

ROC curves for cross-validation and test of the best pipeline for dataset D3

Fig. 5

Learning curves of the best pipeline for dataset D0

Fig. 6

Learning curves of the best pipeline for dataset D1

Fig. 7

Learning curves of the best pipeline for dataset D3

Fig. 8

Best pipelines results in cross-validation and test at different time-points

Fig. 9

Stability scores and feature weights of the best pipeline for dataset D0

Fig. 10

Stability scores and feature weights of the best pipeline for dataset D1

Fig. 11

Stability scores and feature weights of the best pipeline for dataset D3