Every day, scientific journals treat us to another discovery powered by artificial intelligence. Some of these discoveries make sense, but most seem unrealistic. This study, for example, puts the words artificial intelligence, diabetes and mobile phone side by side. It looks like a salad tossed together for the internet's algorithms.
Researchers at Klick Labs carried out the study, published in the IEEE Journal of Biomedical and Health Informatics and covered in the news in October 2026. They used voice recordings made on smartphones of people reading short English sentences chosen at random. The researchers analysed 14,759 phone recordings from 461 adults and turned them into sound spectrograms to look for complex patterns linked to type 2 diabetes. Using the voice alone, and analysing men and women separately, they reached an accuracy of 71.3% in women and 72.4% in men.
The idea came from an earlier study, published in 2023, which suggested that diabetes affects the voice in pitch, intensity, shimmer and jitter, among other things. The new study did not reach the level of the old one, which reported an accuracy of up to 89%. So we can hardly claim to be impressed by this achievement.
Like
Love
Angry
Dislike
