AI that analyzes voice to detect type 2 diabetes: the promising and unexpected study to screen for the disease

Analyser la voix pour diagnostiquer le diabète de type 2

Une étude, présentée lors de la réunion annuelle de l'Association européenne pour l'étude du diabète (EASD) à Madrid, a récemment révélé une méthode inattendue pour dépister le diabète de type 2. Analyser la voix aiderait à détecter les cas non diagnostiqués.

Type 2 diabetes is diagnosed by taking a blood test on an empty stomach to measure blood sugar levels. These laboratory tests are time-consuming and expensive.

Less than 30 seconds

That's why researchers at the Luxembourg Institute of Health have developed an artificial intelligence (AI) model that can tell whether or not a person has type 2 diabetes, and this in… 25 seconds.

In detail, the scientists analyzed voice recordings from 607 adults. Each read a few sentences from their smartphone or laptop. The AI ​​algorithm then looked at various voice characteristics, such as changes in pitch, intensity, and timbre, combined with health data such as age, body mass index (BMI), and blood pressure. This was to identify differences between individuals with and without diabetes. Result: this model was 66% accurate for women and 71% accurate for men.

In France, 800,000 people are unaware that they are diabetic

Nearly half of diabetic adults (around 240 million worldwide) do not know that they suffer from this disease. And for good reason, type 2 diabetes often progresses silently (without any symptoms) for several years.

But early detection and treatment can help prevent serious complications. Diabetes does indeed cause serious long-term complications. The disease accelerates atherosclerosis, leading to myocardial infarction, stroke, and lower limb arteritis.

“Although our results are promising, further research and validation are needed before this approach has the potential to become a first-line diabetes screening strategy“, the authors state.