A funhouse mirror stretches your reflection: near the edges, your body looks longer than it really is. Machine learning models make a similar mistake when asked to guess biological age from blood tests or brain scans. They underestimate age in older people and overestimate it in younger ones, because they are trained on chronological age, which does not match true biological age. True age is hidden, and the model shrinks its predictions toward the mean, like a rubber band.
The study authors came up with a method that calibrates the model using two points, as if adjusting a mirror so it no longer distorts proportions. As a result, the slope came close to one, and the bias disappeared. A check on real data showed that before correction, the models claimed that people with better cognitive function had brains that aged faster, while after correction, the opposite was true. The same goes for kidney function: the link changed sign. This means that many earlier conclusions about aging could have been an artifact rather than the truth.
In the brain age case, biased models claimed that people with better cognitive function have brains that age faster, while unbiased ones said the opposite. The same dataset led to opposite medical recommendations depending on whether the bias was corrected.
In the series Altered Carbon, biological age can be separated from chronological age with technology, but even there estimates of cerebral cortex age suffer from systematic errors if the algorithm is not calibrated. The work reminds us that before trusting 'aging clocks', we need to make sure they do not lie at the edges of the range.
Not published on arXiv. Here are its page, the PDF and the formalization: you can check it yourself, not only read our account.