0
0
Madrid, Spain – Voice analysis, paired with artificial intelligence (AI), could become a groundbreaking tool for detecting undiagnosed cases of type 2 diabetes (T2D), according to new research presented at the Annual Meeting of the European Association for the Study of Diabetes (EASD), held in Madrid from September 9–13. The study suggests that a brief voice recording, combined with basic health data, could help identify individuals with T2D, offering a non-invasive, cost-effective alternative to traditional diabetes screening methods.
Researchers from the Luxembourg Institute of Health developed an AI-based algorithm that uses short voice recordings—an average of just 25 seconds—along with details such as age, sex, body mass index (BMI), and hypertension status. The AI model was able to distinguish whether a person had T2D with 66% accuracy in women and 71% accuracy in men.
Lead author Abir Elbeji emphasized the accessibility of this approach, explaining, “Most current methods of screening for type 2 diabetes require a lot of time and are invasive, lab-based, and costly. Combining AI with voice technology has the potential to make testing more accessible by removing these obstacles. This study is the first step towards using voice analysis as a first-line, highly scalable type 2 diabetes screening strategy.”
Undiagnosed Diabetes: A Global Challenge
Diabetes remains a significant public health issue worldwide, with approximately half of the 240 million adults with diabetes unaware that they have the condition. The majority—around 90%—have T2D. Early detection and treatment are critical to preventing complications such as heart disease, kidney failure, and vision loss, making it crucial to develop innovative screening methods.
The symptoms of T2D can be subtle or non-existent, leading to many cases going undetected. The AI-based voice analysis tool has the potential to revolutionize screening by offering a fast, simple, and scalable method for identifying at-risk individuals.
The Study
To develop and assess the AI model, researchers asked 607 adults from the Colive Voice study, both with and without T2D, to provide voice recordings of themselves reading a few sentences using a smartphone or laptop. The AI then analyzed vocal features—such as pitch, intensity, and tone—to identify differences between those with and without T2D.
The analysis used two advanced techniques: one that captured up to 6,000 vocal characteristics, and a more refined deep-learning approach that focused on 1,024 key features. These features were then compared to known diabetes risk factors such as age, BMI, and hypertension, as well as the American Diabetes Association (ADA) risk assessment tool.
The voice-based AI performed particularly well in women over 60 and in individuals with hypertension, showing strong predictive capacity across risk factors. In addition, the voice analysis achieved a 93% agreement with the ADA risk score, indicating that it could serve as a reliable alternative to questionnaire-based tools.
Next Steps
While the results are promising, the researchers caution that more work is needed before the tool can be implemented in clinical settings. Co-author Dr. Guy Fagherazzi explained, “Further research and validation are necessary before the approach has the potential to become a first-line diabetes screening strategy and help reduce the number of people with undiagnosed type 2 diabetes. Our next steps are to specifically target early-stage type 2 diabetes cases and pre-diabetes.”
The study’s findings highlight the exciting possibilities of AI in healthcare, paving the way for more accessible and efficient methods of disease detection and prevention. If successful, voice analysis could play a pivotal role in tackling the growing global diabetes epidemic.
