AI-powered voice analysis tool could detect Type 2 diabetes

Ten-second voice recordings interpreted by artificial intelligence could be used to screen for Type 2 diabetes, a study has found.

The model developed by the scientists at Klick Labs was 89 per cent accurate for women and 86 per cent accurate for men.

The study suggests that Type 2 diabetes induces subtle changes in a person's voice.

These changes, such as variations in pitch and intensity, are not easily perceivable by the human ear but can be detected and analysed through specialised technology.

This research suggests a potential shift towards more accessible and cost-effective preliminary screening methods for Type 2 diabetes, a condition that remains undiagnosed in almost half of the 240 million adults affected globally.

In the study published in Mayo Clinic Proceedings: Digital Health, participants, both non-diabetic and those with type 2 diabetes, were asked to record a phrase six times daily for two weeks using their smartphones.

A total of 267 participants contributed to more than 18,000 recordings, which were then analysed for differences in 14 vocal features, including pitch and intensity.

The AI model used these vocal features, alongside basic health data such as age, sex, height and weight, to distinguish between non-diabetic individuals and those with Type 2 diabetes.

Jaycee Kaufman, a research scientist at Klick Labs and the first author of the study, said: “Our research highlights significant vocal variations between individuals with and without Type 2 diabetes."

Yan Fossat, vice president of Klick Labs, also highlighted the study’s implications, suggesting that the new approach could “revolutionise healthcare practices as an accessible and affordable digital screening tool”.

While the findings are encouraging, the researchers plan to conduct further studies to confirm their results and to explore the applicability of voice analysis in diagnosing other conditions, such as prediabetes and hypertension.

Additional research involving more extensive and diverse participant groups is considered necessary to validate voice analysis as a universally effective and reliable screening tool for Type 2 diabetes.