Accurate modeling of biological age has clinical value for risk stratification, personalized prevention, and intervention planning, promoting proactive healthcare for aging and age-related diseases. Because aging is multidimensional, robust and interpretable multimodal approaches are needed. We studied 908 non-dementia older adults (> 60 years), collecting data on gait, eye movements, resting-state functional connectivity (rs-FC), and plasma biomarkers (neurofilament light chain, NfL, and glial fibrillary acidic protein, GFAP). Fourteen gait features, two eye movement features, 19 rs-FC features, and plasma GFAP levels were significantly correlated with age ( p < 0.05). Among single-domain models, eye movement features showed the strongest predictive performance (R ² = 0.606; MAE = 3.060). A combined multimodal model achieved markedly higher accuracy (R ² = 0.814; MAE = 1.902). These findings demonstrate that integrating physiological, neurological, and biomarker data substantially improves biological age modeling, supporting the development of comprehensive frameworks to assess aging better and guide timely, targeted preventive strategies.