{
  "id": 10544017,
  "title": "Just 20 seconds of speech could help detect type 2 diabetes using AI-based tool",
  "url": "https://urgent.news/2026/09/28/just-20-seconds-of-speech-could-help-detect-type-2-diabetes-using-ai",
  "topic": "ai",
  "section": "AI",
  "published": "2026-09-28T22:00:06.000Z",
  "source": {
    "name": "Medical Xpress",
    "slug": "medical-xpress",
    "url": "https://medicalxpress.com/news/2026-09-seconds-speech-diabetes-ai-based.html"
  },
  "original_language": "en",
  "account": "Researchers have developed an AI-based tool that can detect type 2 diabetes from just 20 seconds of a person's speech, according to a study being presented at the European Association for the Study of Diabetes meeting in Milan, Italy. The study, which is the largest of its kind, suggests that AI analysis of short voice recordings could serve as a fast, noninvasive and scalable screening method for the condition. Type 2 diabetes is a growing concern, with early detection crucial to prevent complications like heart disease and nerve damage. Currently, screenings involve blood tests or doctor appointments, with fewer than half of eligible adults attending these appointments in the U.K. Previous research has linked type 2 diabetes to speech changes, such as hoarseness, roughness, and reduced breath and voice control while speaking. The researchers, from thymia and RMIT University, trained their AI model on 63,283 voice samples from 21,129 people in the U.K. and U.S., who had reported their diabetes status. The model performed well across different demographics but had lower accuracy for Black participants, those with heart disease, high blood pressure, or obesity, likely due to the overlap of these conditions with type 2 diabetes. In the second evaluation, the speech model correctly identified 75% of people with type 2 diabetes based on their HbA1c test results, with an 82% sensitivity rate and a 47% false-positive rate. The model can also classify people into low, medium, and high risk of type 2 diabetes based on HbA1c results. While the tool shows promise, further clinical validation is required before it can be used to guide blood testing for diabetes screening.",
  "summary": "Signs of type 2 diabetes can be detected in just a few seconds from the way someone talks, according to research being presented at the annual meeting of the European Association for the Study of Diabetes (EASD) in Milan, Italy, Sept. 28–Oct. 2.",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}