Canadian AI Can Predict Type 2 Diabetes from Patient’s Voice
Canadian researchers have successfully trained a machine-learning AI to accurately predict Type 2 diabetes using just six to 10 seconds of a patient’s spoken voice. The model identified 14 acoustic features that distinguish between non-diabetic and Type 2 diabetic individuals. By focusing on vocal features such as changes in pitch and vocal intensity, along with basic health information like age, sex, height, and weight, the AI achieved a diagnosis accuracy of 89 percent for women and 86 percent for men.
The development of this AI model has the potential to significantly cut costs for individuals suffering from the chronic condition. Traditionally, patients are required to undergo in-person diagnostic tests, which can be time-consuming and expensive. By using voice technology, this AI offers a remote and automated diagnosis, removing barriers and enabling more accessible screening for diabetes.
Type 2 diabetes is a condition characterized by high blood sugar levels. It affects more than 4 million people in the UK and is associated with being overweight and having a family history of the disease. The body’s improper reaction to insulin, the hormone that controls sugar absorption in the blood, leads to difficulty in regulating glucose levels. Excess fat in the liver increases the risk of developing Type 2 diabetes.
Symptoms of Type 2 diabetes include fatigue, excessive thirst, frequent urination, and can lead to complications related to nerves, vision, and heart problems. Treatment generally involves lifestyle changes and dietary adjustments, but more severe cases may require medication.
The AI developed by Canadian firm Klick Labs aims to address the issue of undiagnosed individuals with diabetes. It is estimated that nearly half of adults living with diabetes worldwide are unaware of their condition. By utilizing vocal variations, the AI accurately determines the presence of Type 2 diabetes. The research hopes to transform how diabetes is screened within the medical community.
The AI model was trained using recordings from 267 test subjects, with approximately 72 percent of them having been previously diagnosed as nondiabetic. From 18,000 individual recordings, the AI identified 14 acoustic features that enabled accurate predictions of individuals with and without Type 2 diabetes. Notably, the AI’s accuracy improved when the age and body mass index (BMI) of the person were incorporated into the prediction model.
Yan Fossat, the vice president of Klick Labs and the study’s principal investigator, emphasized the potential of voice technology in identifying Type 2 diabetes and other health conditions. He hopes that the non-intrusive and accessible AI approach developed by Klick Labs will aid the millions of undiagnosed individuals silently suffering from diabetes. The next steps involve replicating the study’s findings and potentially expanding voice-diagnosing research to other medical areas.
This AI presents an exciting advancement in the healthcare industry, with the potential to revolutionize practices as an affordable and accessible digital screening tool. By utilizing voice technology, diagnosing Type 2 diabetes can become more convenient and efficient, allowing for early intervention and improved patient outcomes.