Audio analysis enables instant diabetes detection
Thymia, in collaboration with RMIT University, has developed an AI-based tool capable of detecting indicators of Type 2 diabetes within seconds by analyzing a voice recording no longer than 20 seconds. The model uses vocal features such as hoarseness and difficulty controlling breath. It was trained on more than 63,000 voice samples from 21,000 individuals and achieved up to 80% accuracy in identifying risk levels in preliminary tests. This non-invasive screening aims to assist people who do not undergo regular check-ups, serving as a helpful tool before conducting final blood tests, while further validation in clinical settings and performance assessment across different population groups are needed.
Size: 120%Text Size
Notice: This Is an AI-Generated Summary
Comments (0)