StackAge is a Multi-Omics Aging Clock
Producing new aging clocks is easier than overcoming the hurdles to the practical use of existing aging clocks, so the research community is generating new clocks at a fair pace while failing to make much concrete progress on the challenging problem of how to use clocks to assess novel potential rejuvenation therapies. An aging clock measures some combination of parameters that at least appears to reflect biological age. Given that clocks are reverse engineered from epidemiological data via machine learning techniques and the research community has not established clear links between biological age and any of the specific parameters used in a clock, it is entirely unclear as to whether any given clock will accurately reflect the outcome of an actual rejuvenation therapy. Will it understate or overstate the effects of repairing some form of cell and tissue damage? Will its predictions regarding mortality risk turn out to be correct? They only way to find out at present is qualify a specific clock for a specific intervention via the slow and expensive life span studies that everyone wants to avoid. Some way to fix this present situation is needed, and building more clocks seems unlikely to achieve that goal.
Accurate quantification of biological age is essential for early risk stratification and intervention of chronic diseases. Here, we present StackAge, an ensemble-based biological aging clock that integrates large-scale plasma proteomic and metabolomic profiles from 30,376 participants in the UK Biobank. StackAge demonstrated high accuracy in age prediction (Pearson r ≈ 0.93 with chronological age) and substantially enhanced risk prediction for 12 chronic diseases, achieving area under the curve (AUC) exceeding 0.90 for type 2 diabetes, Alzheimer's disease, and chronic kidney disease. Notably, the incorporation of estimated aging rates consistently improved disease prediction beyond conventional omics and demographic features.
Feature interpretation and pathway enrichment analyses revealed that aging-associated biomarkers were enriched in inflammation, metabolic stress, and extracellular matrix remodeling pathways. Mediation analysis further indicated that modifiable lifestyle factors may accelerate biological aging, thereby increasing susceptibility to cardiovascular, neurological, immune, and musculoskeletal disorders. Together, these findings establish a robust multi-omics framework for quantifying individual aging trajectories and highlight biological age as a clinically actionable indicator for precision prevention and health management of age-related diseases.