There’s an important difference between biological age and chronological age. People can be ‘older’ or ‘younger’ biologically, which affects life expectancy. Machine learning can help to assess this difference.
Life expectancy varies and there are marked differences between people and communities. Researchers have developed a methodology to analyze skin cells. Trials based on assessing more than 100 people of different ages showed that it was possible to detect molecular signatures that change as people get older.
For different countries in the world, life expectancy results are published. There are differences between countries and within, with regional variations. Many of these differences are connected to health and poverty. For example, life expectancy in Japan is 74.9 years; in Canada it is 72.3 years; the U.K., 71.4 years; and in the U.S. it is 69.1 years, based on World Health Organization data. As an example of regional variations, in the U.K. with Scotland, the majority of local areas were in the fifth of local areas in the U.K. with the lowest male and female life expectancy at birth (based on Office of National Statistics data). There are multiple factors that explain variations in regional life expectancy, such as poverty, diet, unemployment, deprivation among older people and gender. These factors came to lights in 1980 when the Black Report was published in the U.K. (seen as criticism of the Thatcher government and neo-liberal economics) and they have been verified multiple times ever since.
But what with individual health outcomes and life expectancy? How can the differences between biological age and chronological age be differentiated? A 60 year old man may exhibit signs of old-age whereas an 80 year old woman may show all the signs of being fit and healthy.
To understand this difference, researchers from the Salk Institute (San Diego, California, U.S.) analyzed samples of skin to look fro molecular signatures and then deployed machine learning to assess the results. With the samples, the scientists focused on a type of skin cell called dermal fibroblasts. These cells generate connective tissue and help the skin to heal after injury.
By deploying custom machine-learning algorithms to sort out the data, the researchers discovered key biological markers that indicate aging. Through this the machine learning platform was able to predict a person’s age with less than eight years error on average.
The new research has been published in the journal Genome Biology. The research paper is titled “Predicting age from the transcriptome of human dermal fibroblasts.”