Insomnia is associated with altered arachidonic acid–induced platelet aggregation and lower cholesterol independent of age in aspirin-treated unstable angina: a machine learning analysis

BACKGROUND: In aspirin-treated unstable angina (UA), heterogeneity in on-treatment arachidonic acid (AA)-induced platelet aggregation and cholesterol reduction contributes to residual ischemic risk. Insomnia is prevalent in coronary disease, yet its association with specific aspirin-sensitive platelet aggregation kinetics and lipid parameters remains undefined. METHODS: In a cross-sectional cohort of 198 aspirin-treated UA patients, 38 routine hematologic or biochemical variables including AA-induced platelet aggregation indices were analyzed using principal component analysis (PCA) and unsupervised phenotyping (k-means and hierarchical clustering). Elastic-net logistic regression and machine learning classifiers were used for multivariable modeling, and performance was estimated by cross-validation. Correlation-network analysis and Shapley additive explanations (SHAP) supported interpretability. RESULTS: < .001). Inflammatory markers, including neutrophil count and monocyte count, were no longer selected. Network analysis placed insomnia within the platelet-lipid module linked to PMaxAggR, CHOL, and some other platelet or lipid indices. CONCLUSIONS: In age-matched, aspirin-treated UA patients, insomnia is independently associated with reduced PMaxAggR and lower CHOL, with inflammatory markers no longer significant after age-balancing. These findings delineate an age-independent platelet-cholesterol axis linked to insomnia, potentially reflecting shared autonomic and neurobiological dysregulation, and motivate sleep evaluation as part of platelet-reactivity phenotyping in this population.