Aging is often described as a gradual loss of function, but the real picture is more complex: periods of development, puberty, and later life each have their own physiological patterns. Previously, scientists built biological age models that reduce everything to a single number, but these do not show how the body actually changes at different stages. This work stands apart: it does not predict an individual's age, but reconstructs the population trajectory in a multidimensional space of biomarkers. Nearby in the archive are papers on cellular aging, DNA methylation, and epithelial tissues, but they examine different scales, from molecular clocks to tissue transitions. Here, the focus is on the systemic physiology of the whole organism, which makes it possible to see the overall architecture of the life course.
The authors used data from 11,124 NHANES participants (the National Health and Nutrition Examination Survey in the US) aged 12 to 80. For each sex and each age, they averaged the values of 54 blood biomarkers to obtain stable profiles. They then applied principal component analysis (PCA) to these profiles, which identified two main axes of variation. The first axis was interpreted as sex, the second as age. To estimate the rate of change, they introduced the ETD (Euclidean Trajectory Distance) metric, a weighted distance traveled in the principal component space per year. To check reproducibility, they repeated the analysis on the Korean KNHANES cohort with a smaller set of biomarkers. They also compared the trajectories of the general population and a subgroup without chronic diseases, and studied the influence of obesity, smoking, and depression.
The two main components explained 54% of all biomarker variation. The first axis (PC1) reflected sex differences: it diverged in men and women during adolescence, peaked at ages 20–40, and converged toward old age. The second axis (PC2) described age-related changes and had a nonlinear shape with breakpoints. Based on the distance between the male and female trajectories, they identified five stages: early adolescence, sexual divergence, sexual convergence, late adulthood, and advanced aging. The stage boundaries were determined with narrow confidence intervals, for example, the onset of sexual convergence at 45.3 years (95% CI 42.9–47.6). During the period of maximum dimorphism, men had higher testosterone, hemoglobin, and creatinine, while women had higher SHBG, HDL cholesterol, and platelets. The rate of change (ETD) showed a sharp acceleration in women at ages 40–60, coinciding with menopause, a slight acceleration in both sexes at ages 60–65, and a stronger one after age 75. In healthy people without chronic diseases, the trajectory is shifted toward a younger age: by age 79, the gap reaches 16 years compared with the general population. Obesity, smoking, and depression also shift the trajectory, but in different ways: obesity begins to act from adolescence, smoking from early adulthood, and depression has a stronger effect on women in middle age.
The work shows that physiological aging is not a monotonic process but a sequence of phases with different rates. Unlike neighboring studies that examine individual molecular mechanisms or cellular transitions, this one offers a population-level view of the whole organism. This makes it possible to quantitatively compare the health of different groups and assess how much disease or lifestyle accelerates movement along the life trajectory. The method does not require longitudinal data and can be applied to any cross-sectional samples with biomarkers.
The next step is validation on longitudinal cohorts, where the same people are followed for years. This will make it possible to test whether the population trajectory reflects individual changes. It would also be promising to link the PCLT phases to molecular markers of aging, such as epigenetic clocks, and to clinical outcomes. Perhaps it will be possible to identify critical periods when interventions are most effective for slowing aging.
The method could be useful in the epidemiology of aging, geriatrics, and preventive medicine for monitoring population health and evaluating the effect of interventions. It also provides a tool for comparing the physiological state of different ethnic and social groups.
Test reproducibility in other cohorts with a wider age range and different sets of biomarkers. Link ETD to longitudinal data on morbidity and mortality to establish the predictive value of the metric.
In the biology of aging, the question remains open as to how to reconcile population trajectories with individual rates of aging. This work offers a bridge between them but does not solve the problem of causality: cross-sectional data cannot distinguish a cohort effect from a true age-related change. It is also unclear what biological mechanisms underlie the identified phases and why women have a pronounced menopausal transition while men have no analogous sharp shift.
By age 79, healthy people lag behind the general population along the physiological trajectory by 16 years, meaning their body looks as if they were 63.
Not published on arXiv. Here are its page, the PDF and the formalization: you can check it yourself, not only read our account.