Every city has a road map. From the way roads connect cities and how wide they are, you can guess how old the network is. That is roughly how a neural network learns to estimate a person's age from a map of connections in the brain.
Instead of cities, the brain has 84 regions, and instead of roads, bundles of nerve fibers. The map of these connections is called the connectome. It is built from diffusion MRI scans, which show how water moves along neural pathways. The neural network looks at the structure of connections and names the person's age.
Scientists trained the model on data from four groups of people. In one group it was off by an average of 4.7 years, in another by 9.75 years. But what mattered more was the difference between predicted and real age. If the model says the brain is 70 years old and the person is 65, then the brain is aging faster than usual.
To remove systematic error, the expected trend was subtracted from the difference, yielding an adjusted gap. It turned out to be linked to white matter integrity. People with a higher gap had lower fractional anisotropy in the hippocampus, a region important for memory. What is more, those whose gap was higher at the start showed a steeper drop in fractional anisotropy later on. A brain that looks older really does wear out faster.
When a model trained on the AD-DECODE cohort was transferred to other cohorts, the age prediction error reached 80 years, meaning the model could make a person almost a century older. After linear recalibration, the error dropped to 8 years, but the ordering of people by age still was not recovered.
The idea of a biological age distinct from chronological age echoes H. G. Wells's The Time Machine, where time flows unevenly for different beings. Here, each person's brain ages at its own pace, and the model tries to capture that individual trajectory from the structure of connections.
Not published on arXiv. Here are its page, the PDF and the formalization: you can check it yourself, not only read our account.