Mathematicians and biologists from the Russian startup Gero have created an algorithm that allows you to accurately estimate a person's biological age using a short questionnaire. Their findings were published in the electronic library bioRxiv.org.
In recent years, scientists have been actively looking for ways to estimate a person's biological age. Scientists understand this term as how old or young the tissues and organs of a particular individual look in comparison with other people from his age category.
The discrepancy between calendar and biological age, according to biologists, will help doctors determine whether to pay special attention to the health of their patients in order to help them live as long as possible.
Initially, scientists calculated the biological age of a person by analyzing the concentration of certain protein molecules in the blood, as well as the structure of the DNA wrapper and many other aspects of the body's work. In recent years, biologists have begun to find hints that the same assessments can be obtained without taking blood samples, but by analyzing the work of organs using MRI, X-rays and other diagnostic systems.
Pyrkov and his colleagues recently learned how to calculate the biological age of a person using artificial intelligence and simple fitness trackers that track the level of physical activity. They came to this conclusion after analyzing data collected while observing the lives of approximately one hundred thousand American nurses and paramedics using machine learning methods.
These successes made mathematicians wonder how much the accuracy of predictions between such simple systems that do not require interference with the body's work, and more complex methods that involve the collection of blood and other tissues, differ.
To do this, they used the same dataset from the NHANES project as when working with fitness trackers, which included not only data on physical activity of American medical professionals, but also their blood samples, dietary habits and results of various surveys.
In this comparison, the scientists relied on one simple principle - the results of predictions of all these methods for calculating bio-age should not differ greatly from the chronological age of nurses and medical workers, and at the same time correctly calculate the probability of their death or the acquisition of strokes, heart attacks and other diseases.
Promotional video:
This analysis helped them find out that the prediction accuracy of algorithms using both biomarkers in the blood and survey data was approximately the same. Based on this idea, the scientists created a neural network that helped them select the most important questions that the authors of NHANES used, and create a simple, yet accurate way to calculate bio-age.
A more complex version of this algorithm, which also takes into account the differences in biomarkers, was able to correctly calculate the average life expectancy of NHANES participants - 79.9 years, which differs from the official WHO statistics for the United States by only six months.
Scientists hope that their approach will not only make such predictions more accessible and accurate, but will also help uncover other factors affecting a person's biological age by analyzing other data sets.
