Imagine knowing you’re at risk for type 2 diabetes without the need for a lab, blood test, or needle prick. Simply speak for 20 seconds, and an artificial intelligence (AI) system analyzes your voice and tells you whether you’re at risk. This may sound like something out of a science fiction film, but new research by scientists has brought it closer to reality.
Researchers from Australia’s RMIT University and technology company Thymia have developed an AI model that can assess your risk of type 2 diabetes based on just 20 seconds of voice recording. Researchers say that diabetes affects the body’s nerves, muscles, and blood vessels, causing subtle changes in voice. These changes are difficult for the human ear to detect, but AI algorithms can recognize them.
The AI model analyzes signals such as voice pitch, speech rate, breathing patterns, and voice quality. Based on these patterns, it generates a score for a person’s potential risk of type 2 diabetes.
To develop this technology, researchers used over 63,000 voice recordings from over 21,000 people in the US and UK. The model was then tested on thousands of participants. In some cases, participants were read a written story, while others were given recordings of normal conversations.
During the research, the AI assigned higher risk scores to those who self-reported having type 2 diabetes. Later, when compared with blood test results from some participants, the model also showed positive results.
Health experts clearly state that this technology is not currently a substitute for blood tests. It was developed as a screening tool to identify people who may need further diabetes testing. If AI identifies a risk factor in a person, they may be advised to undergo an HbA1c or other blood test.
According to the report, the model accurately identified many cases, but it also showed the potential for false positives, i.e., identifying some healthy individuals as at risk. This is why doctors still consider blood tests the most reliable method for confirming diabetes.
Millions of people worldwide are affected by type 2 diabetes, and a large number are unaware of the disease. Initial screening using a 20-second voice recording via a mobile phone or app could facilitate access to healthcare, especially for those who do not receive regular health checkups or live in remote areas.
Researchers believe that in the future, such AI tools could help identify the disease at an early stage, allowing timely treatment.While this technology offers considerable promise, scientists say further studies are needed before it can be adopted on a large scale. According to some reports, this research is still in its early stages and its results are yet to be tested in different populations and health conditions.

