study of sleep time link with mandibular jaw movements on individual shap scores

Increased Respiratory Effort Impacts the Risk of  Hypertension and Type 2 Diabetes in Sleep Apnoea

Dr. Jean-Benoit Martinot

Dr. Jean-Benoit Martinot

Head of the Sleep Laboratory at CHU UCL Namur, St. Elisabeth site, Namur, Belgium. Affiliated to the Institute of Experimental and Clinical Research, UCL Bruxelles Woluwe, Belgium.

Dr. Jean-Benoit Martinot is an internist and respiratory physician who actively contributes to the realms of sleep medicine and lung function testing. He is a certified BASS and ESRS Sleep Specialist. At present, he leads the Sleep Laboratory at CHU UCL Namur, St. Elisabeth site, situated in Namur, Belgium. His affiliation extends to the Institute of Experimental and Clinical Research, UCL Bruxelles Woluwe, Belgium. Dr. Martinot boasts a comprehensive career devoted to the study of respiratory effort during sleep. His diligent efforts have yielded a portfolio of approximately 75 peer-reviewed publications, prominently featured in leading international medical journals dedicated to this domain. Notably, a substantial proportion of his works, exceeding 30 publications, center around the intricate realm of sleep mandibular movements.

Exploring the Link Between Sleep Time with Increased Mandibular Jaw Movements and the Risk of HTN or Type 2 Diabetes in OSA

In this Sleep Science Friday, we are glad to share Dr. Martinot’s recent research regarding the pivotal role of heightened respiratory effort during sleep in assessing the risk factors for hypertension (HTN) and also for type 2 diabetes among patients with obstructive sleep apnea (OSA). We invite you to read the following article written by him for this SSF, based on his recent papers “Respiratory effort during sleep and the rate of prevalent type 2 diabetes in obstructive sleep apnoea” (vide Martinot et al. 2023) and “Respiratory effort during sleep and prevalent hypertension in obstructive sleep apnoea” (vide Martinot et al. 2022).

Study Overview

Polysomnography (PSG) has been the traditional diagnostic tool for Obstructive Sleep Apnea (OSA) for a long time. However, recent advancements in bio-signal computing have introduced alternative solutions using single-point contact sensors and artificial intelligence (AI). One such device, which records Mandibular Jaw Movement (MJM) through a wireless sensor attached to the chin, shows promising results in identifying specific signal patterns associated with SDB.

The MJM is a product of brainstem interconnectivity and plays a critical role in stiffening the pharynx before diaphragmatic contraction to maintain pharyngeal patency. During sleep, the lower jaw displaces in the frequency range of 0.15 to 0.60 Hz, which corresponds to the typical breathing rate. Motion in this range, known as MJM, is associated with respiratory effort. The intensity or magnitude of the MJM can offer valuable insights into the level of effort exerted by an individual to overcome airway obstruction. Additionally, prominent lower jaw displacement is often linked to arousals at the end of obstructive apnea or hypopnea episodes.

mjm signals indicate reliably the degree of respiratory effort - ssf 11-08
Figure 1: The Magnitude of MJM Signals Indicates Reliably the Degree of Respiratory Effort.

This 3-minute fragment shows mandibular jaw movements (MJM) of variable amplitudes at the breathing frequency during successive episodes of obstructive hypopneas. The MJM amplitude informs about the level of respiratory effort, i.e., the muscular effort engaged by the respiratory pump to ventilate. Compared to the gold standard esophageal pressure signal to measure respiratory effort, MJM show concomitant and proportional changes in amplitudes indicating they support similar information over time about the level of respiratory effort. Besides, MJM variations in amplitude are also in perfect concordance with the inspiratory flow limitation depicted by the nasal pressure signal. 

Footnote: SpO2: pulsed O2 saturation ; RIP respiratory inductive plethysmography ; FlowPr : nasal cannulae ;FlowTh: thermal flow ; Gyroscopic mandible rotation ; Accelerometric mandible displacement ; POES : oesophageal pressure ; C4:A1 : EEG derivation.

Automated analysis of MJM signals involves simulating the manual PSG scoring process using machine learning (ML) algorithms to identify patterns related to physiological or pathological events, such as sleep stages, apneas, hypopneas, or respiratory effort-related arousals (RERAs). These classifications can then be translated into clinical scores.

However, the Apnea-Hypopnea Index (AHI) has faced criticism for its limitations in reflecting the clinical aspects of sleep apnea syndrome. As a result, other metrics like hypoxic burden and acute cardiovascular responses have been implemented in clinical decision-making.

The Findings

Recently, two consecutive studies applied explainable ML analysis to establish a significant relationship between respiratory effort burden measured by the automated MJM analysis (REMOV) and the prevalence of comorbidities, such as systemic arterial hypertension and type 2 diabetes. These studies included a large sample of 1128 patients suspected of OSA and provided a classification rule that detected prevalent hypertension (HTN) and type 2 diabetes with high performance (ROC-AUC of 0.88 and 0.93, respectively). Moreover, REMOV was found to be the most important risk factor associated with these two pathologies, even more relevant than standard PSG metrics, including AHI and ODI. (Figure 2)

These findings suggest that, in addition to AHI or hypoxic burden, respiratory effort burden should be considered when treating OSA patients with hypertension or type 2 diabetes, potentially influencing treatment decisions for these patients.

study of sleep time link with mandibular jaw movements on individual shap scores
Figure 2: Unveiling the Impact of Sleep Time with Increased Mandibular Jaw Movements on Individual SHAP Scores: A Revealing Study.

Footnote: SHAP score is a qualitative understanding of the relationship between the prediction of the model and the components of the data instance that the model used to generate that prediction; the blue dots represent subjects with primary snoring and/or low values of AHI.

For further insights on Dr. Martinot’s research, you can read his most recent publication here.

Recent publications from ESRS members

  1. Shaffer et al. (2023). Online Sleep Diaries: Considerations for System Development and Recommendations for Data Management. Sleep.
  2. Baglioni et al. (2023). Interactions between insomnia, sleep duration and emotional processes: An ecological momentary assessment of longitudinal influences combining self-report and physiological measures. J Sleep Res.
  3. Röcken et al. (2023). Peripheral arterial tonometry versus polysomnographyin suspected obstructive sleep apnoea. Eur J Med Res.
  4. Dauvilliers et al. (2023). Oral Orexin Receptor 2 Agonist in Narcolepsy Type 1. N Engl J Med.
  5. DelRosso and Ferri (2023). Sleep quality, not only sleep quantity, for understanding the role of genetics, epigenetics, and sleep in ADHD. Sleep.
  6. Squarcio et al. (2023). Ultrasonic vocalisations during rapid eyemovement sleep in the rat. J Sleep Res.
  7. Berger et al. (2023). Effect ofoxybutynin and reboxetine on obstructive sleep apnea: a randomized, placebo-controlled, double-blind, crossover trial. Sleep.
Are you an ESRS member and have just published an article?  
Want your research to be featured in a Sleep Science Friday publication? 
Or, if just have a good idea for an article / saw something that sparked your interest,