WATCH | Neuroscientists predict ‘brain age’

May 27, 2017 - 11:37 AM
Reuters videograb

Neuroscientists from Imperial College London and the University of Edinburgh say they’ve found a way to measure individuals’ ‘brain age’, potentially allowing doctors to one day give better informed health advice to their patients. Reuters’ Jim Drury reports.

A combination of MRI brain scans and machine learning algorithms could tell you your ‘brain age’.

Researchers say their new technique could one day be used to warn patients if they’re at risk of an early grave. First they combined datasets of 2,000 MRI scans with their algorithm. They then applied the technique to members of the Lothian Birth Cohort 1936 – a large group of adults who all had brain scans aged 73.

“We made a prediction of how old their brain appears in reference to this model. And then we compared that prediction to their actual chronological age and we looked at the difference between the chronological age and the brain predicted age. And the idea is that the older your brain predicted age, the older the discrepancy, then the more negative that’s probably going to be for your health,” Dr James Cole, lead researcher, Imperial College London, saying.

Our brains change as we get older, for instance by getting smaller. In the Lothian group those with a brain age older than their actual age performed worse on standard physical measures for healthy ageing.

“Investigators get notified when somebody passes away, and approximately 10 percent of these 660 individuals who had the scans at age 73 have died. And there is a relationship between the discrepancy between, so having an older appearing brain and how likely they were to die from lots of different causes,” Cole, saying.

Those deaths weren’t necessarily brain related, but an unhealthy brain often indicates ill heath elsewhere in the body.

The research is at an early stage, but one day scientists hope it will help doctors persuade those with prematurely ageing brains to make lifestyle changes before it’s too late.