Establishment and Validation of a Predictive Model in Female Patients with Obstructive Sleep Apnea

Objective: To develop a noninvasive clinical diagnostic model based on clinical markers for obstructive sleep apnea (OSA) and to verify its predictive efficacy. Methods: A retrospective analysis was conducted on female patients who underwent diagnostic sleep monitoring and had complete medical records from January 2021 to April 2023 at Zhongshan Hospital affiliated with Fudan University. The risk factors were analyzed using LASSO regression and multivariate Logistic regression to construct a nomogram predictive model and evaluate its performance. Finally, the predictive efficacy of the constructed model was compared with that of the STOP-Bang score. Result: A total of 317 female patients were enrolled. Logistic regression analysis revealed that age (OR = 1.045, 95% CI: 1.02–1.072, p < 0.001), snoring (OR = 8.698, 95% CI: 3.439–24.89, p < 0.001), cerebrovascular disease (OR = 28.15, 95% CI: 2.408–931.7, p = 0.025), and Epworth Sleepiness Scale score (OR = 1.217, 95% CI: 1.112–1.348, p < 0.001) were independent risk factors for OSA in females, while insomnia (OR = 0.125, 95% CI: 0.03–0.423, p = 0.002) served as a protective factor. A nomogram predictive model was constructed using the aforementioned independent predictors, exhibiting good discrimination with a C-index of 0.881 (95% CI: 0.84–0.93) in the training cohort and 0.815 (95% CI: 0.73–0.90) in the validation cohort. Comparing the model’s area under the curve with that of the STOP-Bang score, the model’s predictive efficacy was found to be superior to the STOP-Bang score. Conclusions: The nomogram predictive model demonstrates good accuracy, consistency, and clinical utility. It aids doctors in the early identification of high-risk female patients with OSA in clinical practice, enabling timely preventive and interventional measures.