Hematologic biomarkers of aging (HemeAge) and cardiovascular risk: a machine learning analysis in two cohorts

• Machine-learning model developed to estimate biological age from CBC parameters. • Model trained in 53,355 NHANES participants and applied to 109,844 hospital patients. • CBC-derived HemeAge phenotypes is associated mortality independent of demographics and comorbidities. • Accelerated aging showed higher mortality and MACE risk; resilient aging showed lower risk. • CBC-based biological age offers an accessible tool for cardiovascular risk stratification. Chronological age inadequately reflects aging variability and cardiovascular risk. Biological age derived from routine complete blood count (CBC) parameters may provide a more actionable marker. To develop a machine learning model of biological age using CBC data (HemeAge) and evaluate associations with mortality and major adverse cardiovascular events (MACE) in two large cohorts. An XGBoost model was trained on 53,355 NHANES participants (1999–2010) to predict chronological age from CBC parameters. The model was applied to 109,844 Houston Methodist CVD Registry patients, generating "delta age" (predicted minus chronological age). Patients were classified as Resilient (delta < –10), Proportionate (–10 ≤ delta ≤ 10), or Accelerated (delta > 10). Cox models assessed mortality and MACE risk, adjusting for demographics and clinical factors. Red cell distribution width, mean cell volume, and neutrophil count were key age predictors. Accelerated aging associated with increased mortality risk (HR 3.05, 95% CI 2.41–3.85) and MACE (HR 1.37, 95% CI 1.24–1.51) versus proportionate aging. Resilient aging conferred reduced risk for mortality (HR 0.59, 95% CI 0.52–0.68) and MACE (HR 0.76, 95% CI 0.72–0.81). Associations were strongest in midlife (ages 40–80) and for death and heart failure outcomes and persisted across age-stratified and continuous models. HemeAge independently predicts mortality and cardiovascular risk beyond chronological age. These accessible hematologic markers may enhance risk stratification and inform targeted prevention strategies.