Moderate-severe OSA screening based on support vector machine of the Chinese population facio-cervical measurements dataset: A cross-sectional study

Objectives Obstructive sleep apnea (OSA) has received much
attention as a risk factor for perioperative complications and 68.5% of
OSA patients remain undiagnosed before surgery. Facio-cervical
characteristics may screen OSA for Asians due to smaller upper airways
compared to Caucasians. Thus, our study aimed to explore a
machine-learning model to screen moderate-severe OSA based on
facio-cervical and anthropometric measurements. Design A
cross-sectional study. Setting Data were collected from the
Shanghai Jiao Tong University School of Medicine affiliated Ruijin
Hospital between February 2019 and August 2020. Participants A total of
481 Chinese participants were included in the study. Primary and secondary
outcome (1) Identification of moderate-severe OSA with apnea-hypopnea
index (AHI)15 events·h−1. (2) Verification of the machine learning model.
Results The SABIHC2 model (Sex-Age-Body mass index-maximum Interincisal
distance-ratio of Height to thyro-sternum distance-neck
Circumference-waist Circumference) was set up. The SABIHC2 model could
screen moderate-severe OSA with an area under the curve (AUC)=0.832, the
sensitivity of 0.916, and specificity of 0.749, and performed better than
the STOP-BANG questionnaire, which showed AUC=0.631, the sensitivity of
0.487, and specificity of 0.772. Especially for asymptomatic patients (ESS
< 10), the SABIHC2 model
demonstrated better predictive ability compared
to the STOP-BANG questionnaire, with AUC (0.824 vs. 0.530),
sensitivity (0.892 vs. 0.348), and specificity (0.755 vs. 0.809).
Conclusion The SABIHC2 machine learning model provides a simple and
accurate assessment of moderate-severe OSA in the Chinese population,
especially for those without significant daytime sleepiness.