A hematology-based clock derived from the Study of Longitudinal Aging in Mice to estimate biological age

Biological clocks and other molecular biomarkers of aging are difficult to implement widely in a clinical setting. In this study, we used routinely collected hematological markers to develop an aging clock to predict blood age and determine whether the difference between predicted age and chronologic age (aging gap) is associated with advanced aging in mice. Data from 2,562 mice of both sexes and three strains were drawn from two longitudinal studies of aging. Eight hematological variables and two metabolic indices were collected longitudinally (12,010 observations). Blood age was predicted using a deep neural network. Blood age was significantly correlated with chronological age, and aging gap was positively associated with mortality risk and frailty. Platelets were identified as the strongest age predictor by the deep neural network. An aging clock based on routinely collected blood measures has the potential to provide a practical clinical tool to better understand individual variability in the aging process. The authors used deep learning to derive a biological clock based on routine blood markers in mice that distinguishes slow-aging from fast-aging animals. Drawing on data from the NIH Study of Longitudinal Aging in Mice and a study of aging at The Jackson Laboratory, this clock reveals that platelets are key in predicting biological age.