Researchers have developed an artificial intelligence (AI) tool capable of detecting type 2 diabetes by analyzing a person’s voice in just seconds. The technology identifies subtle speech changes associated with the condition, such as hoarseness and altered breath control.

The study involved 7,319 adults from the United Kingdom, each providing 20-second voice recordings. The AI algorithm was designed to assess these samples and assign a risk score indicating the likelihood of type 2 diabetes. Results showed that the tool gave higher risk scores to individuals with the disease 80 percent of the time. Additionally, among 801 participants who subsequently underwent standard blood tests for diabetes, the AI correctly indicated a higher risk in 75 percent of cases.

Giedre Cepukaityte, who will present the findings at the upcoming European Association for the Study of Diabetes conference, emphasized the practical applications of this approach. According to Cepukaityte, voice samples could be easily collected remotely via phone calls or mobile applications, potentially enabling rapid, non-invasive screening.

While traditional diabetes testing requires blood samples and laboratory analysis, this voice-based AI method could offer a convenient preliminary screening tool, especially in settings where access to medical facilities is limited. However, the researchers note that further validation and refinement are necessary before the technology can be widely implemented in clinical practice.