Posted on 17 May 2021
Longevity briefs provides a short summary of novel research in biology, medicine, or biotechnology that caught the attention of our researchers in Oxford, due to its potential to improve our health, wellbeing, and longevity.
Why is this research important: Scientists usually attempt to quantify the rate of biological ageing – that is to say, how rapidly a person’s cells and tissues are ageing – by measuring factors in the blood or changes to the DNA. Such measurements are obviously not routinely available to the average person. Non-invasive, easily accessible ways of accurately measuring biological ageing could be useful, potentially allowing people to adjust their lifestyle and take precautions for the future based on their rate of ageing.
What did the researchers do: Here, researchers trained a deep neural network to estimate rates of biological ageing and disease risk using motion data from wearable/carriable devices, such as smart watches and smartphones. The data used to train the neural network included week-long steps per minute recordings for nearly 100 000 UK Biobank participants, as well as longitudinal data from 1,876 smartphone and 723 smartwatch users. They then tested the ability of their model (called GeroSense) to predict individual health and disease compared with current blood test-based methods.
Key takeaway(s) from this research: GeroSense behaved similarly to models based on blood tests when it came to predicting future disease risk. While it may seem surprising that something as simple as motion sensing data could be similarly effective as a blood test, it’s worth noting that GeroSense is not just using step count as a reporter of physical activity, but is also spotting nuanced patterns of activity that are associated with declining health. For example, the body’s ability to recover from stress is linked to lifespan and is a good reporter of overall health.
Deep longitudinal phenotyping of wearable sensor data reveals independent markers of longevity, stress, and resilience: https://doi.org/10.18632/aging.202816