A pilot study of the ‘PSGCloud’ — A cloud-based care service delivery and sleep disorders diagnosis system. Part I: Sleep structure and arousal analysis

Objectives The high prevalence of sleep disorders has greatly increased the demands for diagnosis and treatment; while the imbalance in geographical distribution of necessary medical resources coupled with lacking of specialist care expertise in China raise the challenges further. This study aims to explore the effectiveness of our proposed innovative PSGCloud system in providing or enhancing primary care institutions with sleep medicine service capabilities. We will introduce the entire study through a series of three research papers: this first one focuses on sleep structure and arousal analysis, and the other two ensuing papers will cover respiratory as well as limb movement issue, respectively. Methods Firstly, the three Sleep Medical Centers involved in the study uploaded their respectively recorded polysomnography (PSG) data to the PSGCloud system. Secondly, the designated PSG technicians ran the cloud-integrated PSG scoring client of the PSGCloud to download and score PSG recordings, while the same technicians also used the standalone PSG systems built-in scoring software to score the data of the corresponding patient to generate two PSG monitoring reports, respectively. Finally, the two corresponding reports for each patient were compared and analyzed in respect of 20 parameters related to sleep structure and arousal analysis as per their differences, correlations, and agreements. Results The dataset in this study included PSG recordings of 30 patients’ (17 male, 13 female, aged 40.8 ± 12.63 years, Body Mass Index 26.37 ± 4.34 kg/m2). For the parameters related to sleep structure, there was no significant difference between the two systems with regard to total sleep time, sleep efficiency, wake time after sleep onset, sleep latency, and so on. For the parameters concerning arousal, there was no statistically significant difference either. Significant correlations (all ICC > 0.8, and most ICC > 0.9, ICC stands for intra-class correlation) were found in parameters of the two systems. The Bland-Altman analysis showed positive agreements between the two systems. Conclusions This study demonstrated clearly that the sleep scoring agent of our proposed PSGCloud system exhibited high agreement with that of the clinically recognized gold standard of the standalone PSG systems’ in terms of sleep structure and arousal analysis parameters within a limited sampling scope. And it would motivate the implementation of a large-scale validation study for the PSGCloud’s sleep care service delivery model.