Jian Tan, Wei Chen, Dan Yu, Tiantian Peng, Cheng Li, Kai Lv

Department of Otorhinolaryngology, the Central Hospital of Wuhan, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, 430014, People’s Republic of China

Correspondence: Jian Tan; Wei Chen, Department of Otorhinolaryngology, the Central Hospital of Wuhan, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, 430014, People’s Republic of China, Email 344328147@qq.com; cw99990@163.com

Purpose: Due to the lack of clear screening guidelines for different populations, identify strategies for obstructive sleep apnea (OSA) in the outpatient population are unclear, a large number of potential OSA outpatients have not been identified in time. The purpose of our study was to evaluate the applicability and accuracy of artificial intelligence sleep screening in outpatients and to provide a reference for OSA screening in different populations.
Methods: A type IV wearable artificial intelligence sleep monitoring (AISM) device was used to screen adults in the sleep clinic of the Sleep Medical Center for OSA screening, and the general demographic data of the patients were collected. The epidemiological characteristics obtained by AISM screening were analysed. The accuracy of the AISM for the diagnosis of OSA was evaluated and compared with that of polysomnography (PSG).
Results: A total of 1492 participants completed all the studies. The data included 1448 cases total, including 1096 male patients and 352 female patients, with 620 of the total patients being overweight (42.82%) and 429 being obese patients (29.63%). The prevalence of males was 78.19%, and that of females was 55.97% (χ 2 = 95.72, P < 0.001). In males, the risk of moderate to severe OSA was 74.21% in obese people, while in females, the risk was 50%. Age, body mass index (BMI) and the oxygen desaturation index (ODI) were positively correlated and negatively correlated with the lowest and mean oxygen saturation. A total of 100 participants completed both PSG and AISM monitoring, and the accuracies of the AISM in diagnosing mild and moderate-to-severe OSA were 94% and 98%, respectively.
Conclusion: The AISM exhibits good accuracy, and the use of an objective and convenient sleep detection device to screen a large sample population of outpatients is feasible. The prevalence of OSA in adults in sleep clinics is high, and age, sex, and BMI are risk factors for OSA.

Keywords: obstructive sleep apnoea, outpatient population, wearable artificial intelligence sleep monitor, epidemiological characteristics