Prediction of the obstruction sites in the upper airway in sleep-disordered breathing based on snoring sound parameters: a systematic review

Sleep architecture as a candidate for phenotyping sleep bruxism: A narrative physiological review

BACKGROUND

Sleep bruxism (SB), an oral behaviour in otherwise healthy individuals, is characterised by frequent rhythmic masticatory muscle activity (RMMA) during sleep. RMMA/SB episodes occur over various sleep stages (N1-N3 and rapid eye movement (REM)), sleep cycles (non-REM to REM), and frequently with microarousals. It currently remains unclear whether these characteristics of sleep architecture are phenotype candidates for the genesis of RMMA/SB.

OBJECTIVES

This narrative review investigated the relationship between sleep architecture and the occurrence of RMMA as a SB phenotype candidate.

METHODS

PubMed research was performed using keywords related to RMMA/SB and sleep architecture.

RESULTS

In non-SB and SB healthy individuals, RMMA episodes were most frequent in the light non-REM sleep stages N1 and N2, particularly during the ascending phase of sleep cycles. The onset of RMMA/SB episodes in healthy individuals was preceded by a physiological arousal sequence of autonomic cardiovascular to cortical activation. It was not possible to extract a consistent sleep architecture pattern in the presence of sleep comorbidities. The lack of standardisation and variability between subject complexified the search for specific sleep architecture phenotype(s).

CONCLUSION

In otherwise healthy individuals, the genesis of RMMA/SB episodes is largely affected by oscillations in the sleep stage and cycle as well as the occurrence of microarousal. Furthermore, a specific sleep architecture pattern cannot be confirmed in the presence of sleep comorbidity. Further studies are needed to delineate sleep architecture phenotype candidate(s) that contribute to the more accurate diagnosis of SB and treatment approaches using standardised and innovative methodologies.

Systematic reviews of machine learning in healthcare: a literature review

INTRODUCTION

The increasing availability of data and computing power has made machine learning (ML) a viable approach to faster, more efficient healthcare delivery.

METHODS

A systematic literature review (SLR) of published SLRs evaluating ML applications in healthcare settings published between1 January 2010 and 27 March 2023 was conducted.

RESULTS

In total 220 SLRs covering 10,462 ML algorithms were reviewed. The main application of AI in medicine related to the clinical prediction and disease prognosis in oncology and neurology with the use of imaging data. Accuracy, specificity, and sensitivity were provided in 56%, 28%, and 25% SLRs respectively. Internal and external validation was reported in 53% and less than 1% of the cases respectively. The most common modeling approach was neural networks (2,454 ML algorithms), followed by support vector machine and random forest/decision trees (1,578 and 1,522 ML algorithms, respectively).

EXPERT OPINION

The review indicated considerable reporting gaps in terms of the ML's performance, both internal and external validation. Greater accessibility to healthcare data for developers can ensure the faster adoption of ML algorithms into clinical practice.

Effects of demographic and sleep-related factors on snoring sound parameters

OBJECTIVE

To investigate the effect of frequently reported between-individual (viz., age, gender, body mass index [BMI], and apnea-hypopnea index [AHI]) and within-individual (viz., sleep stage and sleep position) snoring sound-related factors on snoring sound parameters in temporal, intensity, and frequency domains.

METHODS

This study included 83 adult snorers (mean ± SD age: 42.2 ± 11.3 yrs; male gender: 59%) who underwent an overnight polysomnography (PSG) and simultaneous sound recording, from which a total of 131,745 snoring events were extracted and analyzed. Data on both between-individual and within-individual factors were extracted from the participants' PSG reports.

RESULTS

Gender did not have any significant effect on snoring sound parameters. The fundamental frequency (FF; coefficient = -0.31; P = 0.02) and dominant frequency (DF; coefficient = -12.43; P < 0.01) of snoring sounds decreased with the increase of age, and the second formant increased (coefficient = 22.91; P = 0.02) with the increase of BMI. Severe obstructive sleep apnea (OSA; AHI ≥30 events/hour), non-rapid eye movement sleep stage 3 (N3), and supine position were all associated with more, longer, and louder snoring events (P < 0.05). Supine position was associated with higher FF and DF, and lateral decubitus positions were associated with higher formants.

CONCLUSIONS

Within the limitations of the current patient profile and included factors, AHI was found to have greater effects on snoring sound parameters than the other between-individual factors. The included within-individual factors were found to have greater effects on snoring sound parameters than the between-individual factors under study.

The role of artificial intelligence in the treatment of obstructive sleep apnea

BACKGROUND

The first-line and most common treatment for obstructive sleep apnea is nasal continuous positive airway pressure, which serves as a pneumatic splint to stabilize the upper airway and is effective when used with appropriate adherence. Continuous positive airway pressure compliance rates remain significantly low despite machine improvements and compliance intervention. Other treatment options include oral appliances, myofunctional therapy, and surgery. The aim of this project is to elucidate the role of artificial intelligence within improving the treatment of obstructive sleep apnea.

METHODS

Related publications between 1999 and 2022 were reviewed from PubMed and Embase databases utilizing search terms "artificial intelligence," "machine learning," "obstructive sleep apnea," and "treatment." Both authors independently screened the results by title/abstract then by full text review. 126 non-duplicate articles were screened, 38 articles were included after title and abstract screen and 30 articles were included after full text review. The inclusion criteria are outline in the PICO framework and involved studies focused on artificial intelligence application in guiding and evaluating obstructive sleep apnea treatment. Non-English articles were excluded.

RESULTS

The role of artificial intelligence in the treatment of OSA was categorized into the following sections: Predicting treatment outcomes of various treatment options, Improving/Evaluating treatment, and Personalizing treatment with improving understanding of underlying mechanisms of OSA.

CONCLUSIONS

Artificial intelligence has the capacity to improve the treatment of OSA through predicting outcomes of treatment options, evaluating the treatment the patient is currently utilizing and increasing understanding of the mechanisms that contribute to OSA disease process and physiology. Implementing AI in guiding treatment decisions allows patients to connect with treatment methods that would be most effective on an individual basis.

Association between snoring frequency and male serum testosterone: Findings from the 2015-2016 National Health and Nutrition Examination Survey

BACKGROUND

This study aimed to assess the association between snoring frequency and male serum testosterone levels.

METHODS

We analyzed data from the 2015 to 2016 National Health and Nutrition Examination Survey. Snoring frequency was relied on self-report, and was divided into never, rarely (1-2 nights a week), occasionally (3-4 nights a week), or frequently (5 or more nights a week) groups. Multivariable analysis controlling for age, race, waist circumference, total cholesterol, diabetes, and hypertension was used to evaluate the association between snoring frequency and male serum testosterone. Furthermore, we performed the subgroup analyses stratified by age and waist circumference.

RESULTS

Our analysis included 1900 participants. In the fully adjusted model, only frequent snoring was inversely associated with male serum testosterone (β -0.053, 95% CI -0.101 to -0.006, P = 0.028); According to the subgroup analysis stratified by age, only in 40-59 years group, frequent snoring was inversely associated with male serum testosterone in the fully adjusted model (β -0.113, 95% CI -0.196 to -0.031, P = 0.007). As for the subgroup analysis stratified by waist circumference, our results showed only in the waist circumference ≥102 cm group (abdominal obesity), frequent snoring was inversely associated with male serum testosterone (β -0.133, 95% CI -0.216 to -0.05, P = 0.002).

CONCLUSIONS

Frequent snoring (5 or more nights a week) is inversely associated with male serum testosterone levels, especially in those aged 40-59 years and those with abdominal obesity.

Auditory Property-Based Features and Artificial Neural Network Classifiers for the Automatic Detection of Low-Intensity Snoring/Breathing Episodes

The definitive diagnosis of obstructive sleep apnea syndrome (OSAS) is made using an overnight polysomnography (PSG) test. This test requires that a patient wears multiple measurement sensors during an overnight hospitalization. However, this setup imposes physical constraints and a heavy burden on the patient. Recent studies have reported on another technique for conducting OSAS screening based on snoring/breathing episodes (SBEs) extracted from recorded data acquired by a noncontact microphone. However, SBEs have a high dynamic range and are barely audible at intensities >90 dB. A method is needed to detect SBEs even in low-signal-to-noise-ratio (SNR) environments. Therefore, we developed a method for the automatic detection of low-intensity SBEs using an artificial neural network (ANN). However, when considering its practical use, this method required further improvement in terms of detection accuracy and speed. To accomplish this, we propose in this study a new method to detect low SBEs based on neural activity pattern (NAP)-based cepstral coefficients (NAPCC) and ANN classifiers. Comparison results of the leave-one-out cross-validation demonstrated that our proposed method is superior to previous methods for the classification of SBEs and non-SBEs, even in low-SNR conditions (accuracy: 85.99 ± 5.69% vs. 75.64 ± 18.8%).