A new study published in the reputed Nature scientific journal seems to get closer to finding certain signs in our blood that could tell us more about our lifespan.
Studies about longevity have long been the focus of many scientific endeavors, with some of them trying to assess which are the factors that can increase lifespan, as well as some others going as far as trying to predict what is the maximum number of years humans can reach. According to one provocative study, there is no such limit at all, with humans reaching a so-called ”mortality plateau” in their 80s. But said study already has many detractors, and it is hard to accept the idea that some of us could live forever.
What are the factors that influence lifespan?
We know that there are several factors that could, to some extent, predict lifespan. One that we can’t change is genetics. Others are connected to our lifestyle. The way we eat could make the top of the list, with Japan’s high life expectancy being connected with diet. But it’s also connected to how much we exercise, if and how many medication we take, how much sun exposure we get, our sleep patterns, our stress levels, the risks we take during life etc. There is much to take into account and an algorithm for a long life has yet to be found.
Blood analysis to predict lifespan
What Joris Deelen and 34 other authors have now come up with are the results of a longitudinal study done on 44168 individuals. The results are interesting, with a set of 14 biomarkers that have been identified, that can predict 5- and 10-year mortality with significantly more accuracy than a conventional risk model that has been used before. The 5-year prediction has a 0.837 statistical correlation with the set of markers, while the 10-year prediction reaches 0.830. To put it in layman’s terms, these 14 markers explain 83-84% of the life expectancy model. It may not sound like much, but no model could make 100% predictions. Furthermore, these figures are up to 6% higher than previous models.
This is not the first study of this kind, as in early 2018, Insilico Medicine publicized an even more ambitious project, based on the long-term analysis of blood cells from about 130,000 South Koreans, Canadians and Eastern Europeans. The full study can be found here. To top things off, the set of markers they found to be relevant, coupled with age, gender and race, were used to create an algorithm which is in beta testing and available for free at https://young.ai/. You can take a blood test for several markers, at a facility of your choice, upload the results, add some personal details and demographics and the software will give a fair estimate of your life expectancy within seconds.
The markers that the second study has found to be of the greatest statistical significance are: glucose, albumin, urea, hemoglobin, HDL cholesterol, triglycerides, erythrocytes, cholesterol, sodium, LDL cholesterol. Gender also ranks high, along with these 10 factors. By statistically computing the influence of these factors, along with several others, with lower impact, a reliable algorithm was obtained, which outperformed previous models. Results here were measured using a metric called MAE (the mean absolute error), which was lower than 6 years.
In other words, this predictor can tell you your lifespan and it will usually be only 5-6 years off. Beats the hell out of tarot cards and glass bowl predictions.
So, although no model is perfect, it looks like we are getting closer to near-perfect predictions on life expectancy, based on mere blood tests. And these results are not so important for patients trying to see their futures, as they are for patients that might want to change what is written, by taking active measures to keep cholesterol down, to have sugar in their blood under control and to try, in general, to lead a healthier life.