A sophisticated predictive framework for identifying obstructive sleep apnea syndrome in women with obesity during pregnancy.

OBJECTIVE

To develop a risk prediction model of obstructive sleep apnea syndrome (OSAS) in obese pregnant women in order to facilitate early intervention.

METHODS

We retrospectively analyzed polysomnography confirmed OSAS cases (n = 828) from Hefei Maternal and Child Health Hospital (Jan 1st,2023-Jan 1st,2025). Cases were classified according to the onset of the OSAS during pregnancy: non-occurrence (n = 526) and occurrence (n = 302). Predictive factors were determined using univariate logistic regression (P < 0.1), LASSO regularization (λ = 0.007) and multivariate logistic regression.

RESULTS

Five independent predictors were identified: tongue size, sleep position, neck circumference, Berlin Questionnaire score and Epworth Sleepiness Scale score. The model was found to have an area under the curve (AUC) of 0.811, good calibration (Hosmer-Lemeshow P = 0.275), low average absolute error (0.017), and net clinical benefit across risk thresholds (0.04-0.95).

CONCLUSION

This five-variable model is a reliable predictor of the risk of OSAS in obese pregnant women and is useful for timely prenatal screening, risk stratification, and early management.