RESUMEN
Objectives: The relationships among positional obstructive sleep apnea (POSA), obstructive sleep apnea (OSA), and periodic limb movement during sleep (PLMS) are unclear. We analyzed these relationships according to OSA severity and explored the underlying mechanisms. Methods: We retrospectively reviewed 6,140 eligible participants who underwent full-night diagnostic polysomnography in four clinical centers over a period of 5 years with eventsynchronized analysis. The PLMS index (PLMI) and periodic limb movements with arousal index (PLMAI) were evaluated. The effects of POSA on the PLMI, PLMAI, and PLMS were analyzed according to OSA severity. Results: The mean PLMI and PLMAI, as well as PLMS prevalence, were significantly lower in those with severe OSA than in those with mild and moderate OSA. The mean PLMI was higher in mild OSA group than in control group. The mean PLMI (4.80 ± 12.71 vs. 2.59 ± 9.82 events/h, p < 0.001) and PLMAI (0.89 ± 3.66 vs. 0.53 ± 3.33 events/h, p < 0.001), and the prevalence of PLMS (11% vs. 5.3%, p < 0.001) were higher in patients with POSA than patients with non-POSA. This trend was particularly marked in severe OSA group (OR 1.55, 95%CI [1.07-2.27]) and less so in mild (OR 0.56, 95%CI [0.30-1.03]) and moderate (OR 1.82, 95%CI [0.99-3.34]) OSA groups. Conclusion: The POSA group tended to have a higher prevalence of PLMS, particularly in those with severe OSA. If PLMS is prominent, diagnosis and treatment of POSA and OSA may be considered.
RESUMEN
Although understanding the physiological mechanisms of obstructive sleep apnea (OSA) is important for treating OSA, limited studies have examined OSA patients' sleep architecture at the epoch-by-epoch level and analysed the impact of sleep position and stage on OSA pathogenesis. The epoch-labelled polysomnogram was analysed multidimensionally to investigate the effect of sleep position on the sleep architecture and risk factors of apnea in patients with OSA. This retrospective multicentric case-control study reviewed full-night diagnostic polysomnography of 6983 participants. The difference in the proportion of time spent supine during non-rapid eye movement (NREM) and REM stages, and the mean duration of respiratory events per body position were evaluated. The frequency of sleep stage transition per body position shift type was computed. Further subgroup analysis was performed based on OSA severity and positional dependency. Supine time in patients with OSA varied across sleep stages, with lower proportions in N3 and REM, and shorter durations with severity. Patients with OSA spent less time in supine positions during N3 and REM, and experienced longer apnea events in both positions compared to the control group. The frequency of all sleep stage transitions increased with OSA severity and was higher among non-positional OSA than positional OSA and the control group, regardless of body position shift type. The sleep stage transition from N3 and REM to wakefulness was notably heightened during position shift. Understanding the sleep architecture of patients with OSA requires analysing various sleep characteristics including sleep position simultaneously, with future studies focusing on position detection to predict sleep stages and respiratory events.
RESUMEN
STUDY OBJECTIVES: Polysomnography (PSG) scoring is labor-intensive, subjective, and often ambiguous. Recently several deep learning (DL) models for automated sleep scoring have been developed, they are tied to a fixed amount of input channels and resolution. In this study, we constructed a standardized image-based PSG dataset in order to overcome the heterogeneity of raw signal data obtained from various PSG devices and various sleep laboratory environments. METHODS: All individually exported European data format files containing raw signals were converted into images with an annotation file, which contained the demographics, diagnoses, and sleep statistics. An image-based DL model for automatic sleep staging was developed, compared with a signal-based model, and validated in an external dataset. RESULTS: We constructed 10253 image-based PSG datasets using a standardized format. Among these, 7745 diagnostic PSG data were used to develop our DL model. The DL model using the image dataset showed similar performance to the signal-based dataset for the same subject. The overall DL accuracy was greater than 80%, even with severe obstructive sleep apnea. Moreover, for the first time, we showed explainable DL in the field of sleep medicine as visualized key inference regions using Eigen-class activation maps. Furthermore, when a DL model for sleep scoring performs external validation, we achieved a relatively good performance. CONCLUSIONS: Our main contribution demonstrates the availability of a standardized image-based dataset, and highlights that changing the data sampling rate or number of sensors may not require retraining, although performance decreases slightly as the number of sensors decreases.
