A project shows how important research is in medicine and how AI can help: it involves finding out if someone has type 2 diabetes by analyzing their voice.
A team of researchers from LIH, the Luxembourg Institute of Health, has developed, under the direction of Doctor Guy Fagherazzi, a technology that analyzes voices. In diabetics, there are vocal changes that may not be perceptible to the human ear, but which can be detected using artificial intelligence, explains Dr. Fagherazzi.
“People who have diabetes have a different voice from people who do not have diabetes, and this can be explained by many small factors and different mechanisms. Hyperglycemia in diabetics, fatigue, neuropathies, which can be present in these people and also an overall voice that is a little hoarser when you have diabetes which is present.”
The data was collected via the international Colive Voice study. To date, it has more than 8,000 voice samples in different languages. For the specific analysis of type 2 diabetes, an English voice sample of approximately 600 people from the United States was used. People had to read a text for 20 to 30 seconds, then the recordings were compared by the AI to the base material.
Dr. Fagherazzi: “And so, we were able to show, for example, that people who had type 2 diabetes had very high scores compared to people who did not have diabetes, or it was much lower. And that “That’s what allows us to separate the two groups.”
-It is estimated that around 800 million people worldwide have diabetes, but only half of them are diagnosed. This new technology should inform easily and quickly about the risk of diabetes.
Dr. Fagherazzi: “Today, with voice, we will never replace a diagnosis of diabetes which must be made in a laboratory with a blood test and the assessment of blood sugar. What we want is to detect people who are at high risk of developing diabetes And so today with the technologies that we have developed, we are correct in approximately 75% of cases. hope.”
The more voice samples researchers have, the more precise the tool can become, adds Doctor Fagherazzi. This is why he is calling on volunteers to participate in the study via the website www.colivevoice.org.
Related News :