Machine learning flags sleep disorders in postmenopausal women
In 2,397 postmenopausal women, sleep disorder prevalence reached 62.4%, with depressive status showing the strongest association at an odds ratio of 4.067.
Research Square
In a cross-sectional study of 2,397 naturally postmenopausal women aged 45 to 60 years from China, researchers evaluated machine learning models to screen for sleep disorders. Using data from the 2018 China Health and Retirement Longitudinal Study alongside an external validation group of 137 women recruited from a hospital sleep clinic, the authors examined six machine learning algorithms and logistic regression to identify key risk factors. The overall prevalence of sleep disorders in the primary cohort was 62.4%. Depressive status, defined by a CESD-10 score of 10 or higher, showed the strongest association with sleep disorders, carrying an odds ratio of 4.067 (95% CI 3.345–4.944).
Why it matters
Sleep disturbances are common after menopause and are linked to long-term health risks. Clarifying strong contributors like depressive symptoms can aid the development of targeted screening tools for women undergoing reproductive aging.
Caveats
This study is a preprint and has not yet undergone peer review. Because the data are cross-sectional, the observed associations do not establish causality, and the external validation cohort was limited to 137 clinic patients.
The paper
Third Affiliated Hospital of Sun Yat-sen University
Research Square · 29 Sep 2026 · Preprint, not peer-reviewed