Scientists have developed an AI that can help diagnose type 2 diabetes from changes in a person's voice
Scientists at the Luxembourg University of Health have proposed a new method of detecting type 2 diabetes.
Until now, such a diagnosis has been made based on the results of blood tests and the patient's general condition, but thanks to the capabilities of AI, an unexpected method of diagnosis has been developed.
Here's What We Know
Scientists conducted an experiment, which consisted of the following: 607 people, both diabetic and healthy, took voice samples, asking them to read out a certain text for 25 seconds.
As a result of the Colive Voice study, a specially designed AI analysed voice pitch, intensity and tone to identify differences between people with and without diabetes, taking into account gender, weight and a number of other physical indicators, and produced a result that surprised the scientists. The AI correctly identified diabetes in 66 per cent of women and 71 per cent of men.
The author of the technique, Abir Elbeji, from the Luxembourg Institute of Public Health, commented on the study:
'Combining AI with voice technology could make testing more accessible. This study is a first step towards using voice analysis as a screening strategy for type 2 diabetes.
Unfortunately, about half of adults with diabetes, roughly 240 million people, don't know they have the disease because symptoms may be mild or absent in the early stages. But early detection and timely therapy can save people from a large number of complications.
Of course, this development requires additional research and increased effectiveness, but it is quite possible that a few more years will pass and this practice will become commonplace in clinics around the world.
Go Deeper:
Source: SciTechDaily