Published • loading... • Updated
AI Can Detect Type 2 Diabetes From 20 Seconds of Speech
The model trained on 63,283 voice samples and identified diabetes risk with 80% accuracy in one UK evaluation, researchers said.
- Researchers presenting at the European Association for the Study of Diabetes annual meeting in Milan unveiled an AI tool that screens for type 2 diabetes using 20-second voice recordings.
- Standard diagnosis currently relies on blood tests measuring average blood glucose, yet research shows only 40% of eligible adults attend routine screenings, leaving many cases undiagnosed.
- Engineers at deep tech company thymia and RMIT University in Melbourne, Australia, trained the system on more than 63,000 voice samples, achieving 82% sensitivity in identifying individuals with the condition.
- Performance proved lower among Black participants and individuals with heart disease, as these conditions often cause similar vocal changes; the tool serves as a triage aid, not a diagnostic replacement.
- Dr. Lucy Chambers, head of research impact and communications at Diabetes UK, noted that while AI technologies could help identify more patients, they must be "rigorously designed and tested" to ensure accuracy.
Insights by Ground AI
47 Articles
47 Articles
A recent study showed the possibility of using artificial intelligence sound analysis for the initial detection of type II diabetes indicators, through a short audio recording of not more than 20 seconds.
·Syrian Arab Republic (the)
Read Full Article
Coverage Details
Total News Sources47
Leaning Left8Leaning Right7Center25Last UpdatedBias Distribution63% Center
Bias Distribution
- 63% of the sources are Center
63% Center
L 20%
C 63%
R 17%
Factuality
To view factuality data please Upgrade to Premium
