Machine learning is being used to understand why some people age quicker than others
There is huge variation in how we age. While some 70-year-olds still run marathons, others are living with multiple life-limiting conditions. To better understand why, several “biological clocks” have been developed to identify biological signatures of ageing.
We know that some chemical structures on our DNA change as we age, and the ends of our chromosomes shorten. But how these signatures relate to each other and to the wider process of ageing is not always clear.
Another type of biological clock considers protein expression in our blood. Recently, researchers from Massachusetts General Hospital looked at nearly 3,000 proteins in the blood of 45,000 UK individuals aged between 40 and 70. They then used machine learning to identify 200 of these proteins that could accurately predict chronological age. The results were published in August in Nature Medicine.
While protein and chronological ages tended to be similar for most people in the sample, where numbers diverged the scientists saw differences in future disease and mortality risk. People who had a higher “protein age” than chronological age were at greater risk of 18 chronic diseases, including cardiovascular diseases, diabetes and certain cancers. They were also more susceptible to frailty and cognitive decline. By contrast, those individuals who were the “slowest agers” were less likely to develop diseases such as dementia or Alzheimer’s.
The 200 “ageing proteins” spanned many functional categories, including elastin and collagen, which make up the support structure between cells, and proteins involved in the immune response, reproduction and hormone regulation.
Even more excitingly, the “protein clock” performed as well in the UK cohort as it did in populations from China and Finland. While more work is needed to understand why some people’s protein signature “ages” faster than others’, this breakthrough raises the prospect of developing a single test that could describe a person’s risk of many chronic conditions far more accurately than their date of birth. Such a test could transform preventive healthcare.
This article is from New Humanist’s winter 2024 issue. Subscribe now.
