Of people estimating age is not only impolite, but also often difficult.
But a single image would seem to be enough for artificial intelligence to reliably analyze a person’s age, according to a recent study by Osaka Metropolitan University. In it, the artificial intelligence’s guesses were combined with the actual age of the people depicted with a certainty of 0.95.
Correlation describes how well the AI’s guesses hit the spot. The best possible would be 1, and the closer you get to it, the better the guesses will be. In general, a correlation exceeding 0.9 can be considered strong.
Average the artificial intelligence’s guess about the age was correct by a margin of three years.
However, just a passport photo won’t do the trick. Instead, the Japanese artificial intelligence analyzes lung x-rays taken from healthy people during health checks.
A total of 67,099 photos were viewed from 2008–2021. There were 36,051 individuals to be imaged, i.e. several lung images taken at different stages were used.
So aging can be seen in the lungs quite accurately if you know how to look.
Research in the second part, the AI was shown 34,197 lung images from people suffering from chronic diseases such as COPD or high blood pressure.
Artificial intelligence systematically evaluates lung images of patients as taken from people who are older than they really are.
Therefore, an overly high age estimate given by artificial intelligence may therefore indicate the probability that the person described as healthy may have an as yet undetected disease.
Next, the team plans to continue developing artificial intelligence as a diagnostic tool.
#Artificial #Intelligence #Artificial #intelligence #age #lung #image