Could the sound of your voice hold the key to diagnosing type 2 diabetes? A pioneering study suggests that it might.
Researchers have demonstrated that artificial intelligence (AI) technology can analyze a brief voice recording and detect signs of this common condition, potentially opening the door to a revolutionary, non-invasive screening method.
The Study: A New Frontier in Diabetes Screening
Presented at the annual meeting of the European Association for the Study of Diabetes (EASD), this research tested an AI model’s ability to identify diabetes from voice recordings.
Results showed the model correctly detected diabetes with 66% accuracy in women and 71% in men. While not yet published in a peer-reviewed journal, these findings highlight the potential for innovative tools to address undiagnosed cases.
Guy Fagherazzi, Ph.D., co-author of the study and director of precision health at the Luxembourg Institute of Health, noted, “We have shown that people with diabetes have different voice patterns when compared to similar people without diabetes.”
He added, “We believe that this technology will never be accurate enough to become a diagnostic tool for type 2 diabetes that could replace a blood test. On the other hand, we are strongly convinced that this could one day become an efficient solution to screen for diabetes and identify at-risk individuals or potential undiagnosed cases.”
This approach could significantly impact global health, as over 240 million people worldwide are unaware they have diabetes. Early detection is crucial for managing the disease and preventing complications.
How the Study Was Conducted
The study involved 607 adults, equally split between those with and without diabetes.
Participants provided a 25-second voice recording using their smartphones or laptops, reading a series of sentences aloud. Researchers observed that individuals with diabetes were generally older and more likely to have obesity than those without.
To analyze the recordings, researchers employed two sophisticated methods. The first examined up to 6,000 detailed vocal traits, while the second focused on approximately 1,000 key features using a deep-learning approach. They integrated basic health information such as age, sex, body mass index (BMI), and hypertension status into the AI model.
The results were promising. The algorithm was particularly effective for women over 60 and individuals with hypertension, groups where diabetes often presents a more distinct voice signature.
“Women are usually easier to discriminate using voice when there is a health issue,” said Dr. Fagherazzi. “Hypertension is also known to affect voice parameters, so we can speculate that people who have both diabetes and hypertension have an even more distinguishable voice.”
Why Diabetes Might Alter the Voice
What causes these vocal changes in people with diabetes? Researchers believe the effects of the disease on various physiological systems may play a role. Chronic high blood sugar levels, fatigue, acid reflux, reduced lung capacity, and nerve damage are all potential contributors.
According to Susan Spratt, M.D., an endocrinologist at Duke University School of Medicine, dehydration caused by diabetes could impact vocal cord tissue and the lining of the mouth, altering speech.
“Diabetes can cause dehydration, which can affect vocal cord tissue, as well as the tissue lining the mouth and the tongue,” said Dr. Spratt, who was not involved in the study. This might make speech sound more abrupt or “sticky.” Additionally, long-term diabetes can damage nerves, including those affecting hearing, which could influence speech patterns.
Dr. Fagherazzi emphasized that the differences in voice traits between people with and without diabetes arise from subtle combinations of changes.
“Some people with more than 10 to 20 years of diabetes experience voice changes that can be detectable by the human ear, but overall, it is mostly thanks to the progress of audio signal processing and AI that we can now detect these subtle changes.”
A Promising Complement to Traditional Methods
While the technology shows promise, experts agree that voice-based diagnostics are not ready to replace established methods.
Kevin Peterson, M.D., MPH, vice president of primary care at the American Diabetes Association, stated, “This is a hypothesis-generating study. It points out possibilities to pursue. It is important to determine what effect such technology would have on a ‘real’ population before considering implementation.”
Currently, blood tests and questionnaires remain the standard for diagnosing diabetes. The U.S. Centers for Disease Control and Prevention (CDC) and the American Diabetes Association emphasize the reliability of these traditional tools.
However, the study revealed that the AI method was 93% in agreement with the ADA’s questionnaire-based risk assessment, suggesting its potential as a supplemental screening tool.
The Future of Voice-Based Diagnostics
Voice-based AI screening could revolutionize how diabetes and other diseases are detected. Scientists are already exploring similar applications for conditions such as Parkinson’s disease, heart disease, and depression. As technology evolves, the accuracy of these methods is expected to improve.
Although it may take years of research and validation before voice diagnostics are widely adopted, the potential benefits are immense. A simple, non-invasive test accessible via a smartphone could reach millions of people who might otherwise remain undiagnosed.
Dr. Fagherazzi and his team are optimistic about the technology’s future. “This could significantly reduce the worldwide diabetes burden, as half of the population with diabetes ignores it,” he said.
For now, the study marks an exciting step forward in precision health, demonstrating the power of AI to uncover insights hidden in everyday aspects of human life, like the voice.
While much work remains, the possibility of voice-based diagnostics offers hope for a more accessible and proactive approach to managing chronic diseases like diabetes.
Recommended For You
