Association of the endothelial activation and stress index with cognitive function in older adults: a cross-sectional study with machine learning

Background Age-associated memory impairment (AAMI) is a predementia state linked to endothelial dysfunction. The endothelial activation and stress index (EASIX) quantifies endothelial injury, yet its association with cognitive function remains unvalidated in population studies. This study aimed to evaluate the relationship between EASIX and cognitive performance. Methods Data from adults aged ≥ 60 years in the NHANES 2011–2014 were analyzed. Multiple linear regression assessed associations between EASIX and cognitive function scores. LASSO regression selected variables, and six machine learning models (e.g., Random Forest, XGBoost) and two ensemble strategies were developed. SHAP values interpreted feature importance. Results Among 2,763 participants, EASIX showed a significant negative correlation with all cognitive scores (P < 0.05). The Random Forest model outperformed other models. SHAP analysis identified EASIX as one of the top five influential variables, with cognitive function levels demonstrating a declining trend as EASIX score increased, particularly among older adults. Conclusion EASIX is significantly negatively associated with cognitive function, especially in advanced age. It shows promise as a blood-based biomarker for early screening and risk assessment of cognitive decline, supporting its potential clinical utility.