Machine learning and geometric morphometrics to predict obstructive sleep apnea from 3D craniofacial scans.

BACKGROUND: Obstructive sleep apnea (OSA) remains massively underdiagnosed, due to limited access to polysomnography (PSG), the highly complex gold standard for diagnosis. Performance scores in predicting OSA are evaluated for machine learning (ML) analysis applied to 3D maxillofacial shapes. METHODS: The 3D maxillofacial shapes were scanned on 280 Caucasian men with suspected OSA. All participants underwent single night in-home or in-laboratory sleep testing with PSG (Nox A1, Resmed, Australia), with concomitant 3D scanning (Sense v2, 3D systems corporation, USA). Anthropometric data, comorbidities, medication, BERLIN, and NoSAS questionnaires were also collected at baseline. The PSG recordings were manually scored at the reference sleep center. The 3D craniofacial scans were processed by geometric morphometrics, and 13 different supervised algorithms, varying from simple to more advanced, were trained and tested. Results for OSAS recognition by ML models were then compared with scores for specificity and sensitivity obtained using BERLIN and NoSAS questionnaires. RESULTS: All valid scans (n = 267) were included in the analysis (patient mean age: 59 ± 9 years; BMI: 27 ± 4 kg/m2). For PSG-derived AHI≥15 events/h, the 56% specificity obtained for ML analysis of 3D craniofacial shapes was higher than for the questionnaires (Berlin: 50%; NoSAS: 40%). A sensitivity of 80% was obtained using ML analysis, compared to nearly 90% for NoSAS and 61% for the BERLIN questionnaire. The auROC score was further improved when 3D geometric morphometrics were combined with patient anthropometrics (auROC = 0.75). CONCLUSION: The combination of 3D geometric morphometrics with ML is proposed as a rapid, efficient, and inexpensive screening tool for OSA. TRIAL REGISTRATION NUMBER: NCT03632382; Date of registration: 15-08-2018.

Modeling Patient-Specific Desaturation Patterns in Sleep Apnea.

OBJECTIVE: The physiological mechanisms involved in cardio-respiratory responses to sleep apnea events are not yet fully elucidated. A model-based approach is proposed to analyse the acute desaturation response to obstructive apneas. METHODS: An integrated model of cardio-respiratory interactions was proposed and parameters were identified, using an evolutionary algorithm, on a database composed of 107 obstructive apneas acquired from 10 patients (HYPNOS clinical study). Unsupervised clustering was applied to the identified parameters in order to characterize the phenotype of each response to obstructive apneas. RESULTS: A close match was observed between simulated oxygen saturation ( SaO2) and experimental SaO2 in all identifications (median RMSE = 1.3892 %). Two clusters of parameters, associated with different dynamics related to sleep apnea and periodic breathing were obtained. CONCLUSION AND SIGNIFICANCE: The proposed patient and event-specific model-based analysis provides understanding on specific desaturation patterns, consequent to apnea events, with potential applications for personalized diagnosis and treatment.

Kinesthetic stimulation for obstructive sleep apnea syndrome: An "on-off" proof of concept trial.

Obstructive sleep apnea (OSA) occurs when the upper airway narrows or collapses due to the loss of upper airway muscle activation at sleep onset. This study investigated the effectiveness of triggered kinesthetic stimulation in patients with OSA. This proof-of-concept, open-label, multicenter prospective study was conducted on 24 patients with severe OSA. During a one night evaluation, kinesthetic stimulation was intermittently delivered in 30 minute periods. The duration of apneas and hypopneas during Stim on and Stim off periods were compared. Five hospital-based university centers in France participated. Sleep studies were evaluated by a single scorer at a core laboratory (CHU Grenoble). Results show that during the Stim on phases, statistically significant decreases in durations of apneas and hypopneas were observed in 56% and 46% of patients, respectively. Overall, 75% of patients showed an improvement in apneas or hypopneas durations. The mean reduction in durations for patients with a significant decrease was 4.86 seconds for apneas and 6.00 seconds for hypopneas. This proof of concept study is the first to identify kinesthetic stimulation as a potentially effective therapy for OSA. These data justify evaluation in a controlled study.

Validation of the System One RemStar Auto A-Flex for Obstructive Sleep Apnea Treatment and Detection of Residual Apnea-Hypopnea Index: A European Randomized Trial.

STUDY OBJECTIVES: Autotitrating continuous positive airway pressure (CPAP) devices adjust pressure in response to changes in airflow and are an alternative to attended in-laboratory titration polysomnography (PSG) to determine optimal pressure levels. The aim of this study was to compare the performance of the System One RemStar Auto A-Flex (Philips Respironics, Murrysville, PA, USA) automatically adjusted positive airway pressure (APAP) mode to manually titrated, fixed pressure CPAP and to validate the device's breathing event detection capabilities against attended in-laboratory PSG. METHODS: Sixty-one patients investigated in five centers for moderate to severe obstructive sleep apnea between May 2012 and June 2013 were invited to participate. Participants underwent two full-night attended polysomnograms in random order with manually titrated, fixed pressure CPAP versus APAP. RESULTS: Fifty-three participants with a mean apnea-hypopnea index (AHI) of 45.9 ± 23 completed two sleep studies and were included in the analysis. There were significant but not clinically relevant differences between APAP and CPAP respectively: Apnea index [1.0 (2.8 ± 0.8), median (mean ± standard deviation)] versus [1.8 (5.3 ± 11.5)], p = 0.004; percentage of N1 sleep [12.3 (15.9 ± 0.5)] versus [14.3 (18.9 ± 12.7)], p = 0.028. AHI values differed between PSG [2.8 (5.5. ± 9.3)] and device [3.7 (6.0 ± 8.6)], p = 0.003). Regarding residual events detection, intraclass correlation coefficients for AHI were strong (0.956, p < 0.001) and the area under the curve was 0.988 (AHI cut-off value of 10). CONCLUSIONS: The new APAP modality was effective and residual apnea-hypopnea indices calculated by the device strongly correlated to those assessed by PSG. COMMENTARY: A commentary on this article appears in this issue on page 167.