A new AI-driven diagnostic test for Type 2 diabetes can accurately identify the condition with just a six-second voice clip, according to a study published in Mayo Clinic Proceedings: Digital Health. Developed by researchers from Klick Labs in Toronto, Canada, this technology combines voice analysis with artificial intelligence and offers a convenient and non-intrusive way to screen for diabetes. The test is reported to have an 89% accuracy rate for women and 86% for men.
The process involves recording a short voice sample, along with basic health data like age, sex, height, and weight, into a smartphone. Using this information, an AI model is created to distinguish between individuals with and without Type 2 diabetes. Researchers collected over 18,000 voice recordings from 267 participants over a two-week period and analyzed 14 acoustic features to detect differences in the voices of non-diabetic and Type 2 diabetic individuals. Some of the vocal variations identified, including changes in pitch and intensity, are imperceptible to the human ear.
This innovation addresses the issue of millions of undiagnosed individuals with Type 2 diabetes, as nearly 50% of adults with diabetes worldwide are unaware of their condition. The non-intrusive and accessible nature of the test has the potential to screen a large number of people and may transform diabetes screening practices. It could also be extended to screen for other health conditions, such as high blood pressure, prediabetes, and women’s health, making it an affordable and widely accessible digital screening tool that could revolutionize healthcare practices.
Article written by Neil Shaw| Image by Unsplash