Using Interpretable Machine Learning with SHAP to Assess Dynapenic Abdominal Obesity as a Stroke Risk Predictor: A Prospective Cohort Study.

Background: Stroke is a major cause of mortality and disability worldwide, with a particularly high burden in China. While dynapenic abdominal obesity (DAO) is associated with adverse cardiometabolic outcomes, its relationship with stroke risk remains unclear. We examined whether DAO predicts stroke using interpretable machine learning in a nationally representative cohort of middle-aged and older Chinese adults. Methods: We analysed prospective data from the China Health and Retirement Longitudinal Study, including 11,207 participants aged ≥ 45 years. Dynapenia was defined as a handgrip strength ≤ 28 kg (men)/≤ 18 kg (women); abdominal obesity was defined as a waist circumference ≥ 90 cm (men)/≥ 80 cm (women). Stroke events were identified via self-reported physician diagnoses. We employed logistic regression, subgroup analyses, multiple machine learning models, and Shapley additive explanations (SHAP) to assess the association and evaluate robustness. Results: Over the 4-year follow-up period, 210 (1.9%) participants experienced stroke. DAO was significantly associated with increased stroke risk (adjusted OR = 1.58, 95% CI: 1.21-2.06). Subgroup analysis demonstrated consistent associations across all subgroups (all interaction Conclusion: DAO was independently associated with an increased risk of stroke, with an interpretable machine learning model further supporting its potential as a predictor. Maintaining muscle strength and managing abdominal obesity may reduce the risk of stroke in older adults. These findings suggest that DAO may serve as a potential risk marker for stroke. Future research, including external validation and implementation studies, is needed before any recommendations for screening or intervention can be made.