A new study conducted by Klick Labs uses artificial intelligence and voice technology to identify diabetes in a way that could be as simple as asking someone to say a few phrases into a smartphone. This is a significant advancement in the field of diabetes diagnosis.
The new study describes how scientists created an AI model to determine if a person has Type 2 diabetes using six to ten seconds of the individual’s voice and basic health information like age, sex, height, and weight. The study was published in Mayo Clinic Proceedings: Digital Health. For women, the accuracy rate of the model is 89%, while for males it is 86%.
Researchers at Klick Labs asked 267 participants in the study—who were classified as Type 2 or non-diabetic—to record a word into their smartphone six times a day for a period of two weeks. Scientists examined 14 acoustic characteristics from over 18,000 recordings to look for variations between those without diabetes and those with Type 2 diabetes.
Our research highlights significant vocal variations between individuals with and without Type 2 diabetes and could transform how the medical community screens for diabetes. Current methods of detection can require a lot of time, travel, and cost. Voice technology has the potential to remove these barriers entirely.
Jaycee Kaufman, Study First Author and Research Scientist, Klick
The Klick Labs team examined several aspects of voice, such as pitch and intensity variations that are imperceptible to the human ear. Scientists were able to identify Type 2 diabetes-related alterations in the voice through the use of signal processing. Kaufman noted that the manner of those voice alterations seemed different for men and women, which was surprising.
A Potential New Screening Tool for Undiagnosed Diabetes
According to the International Diabetes Federation, roughly one in every two persons, or 240 million adults globally, are unaware they have diabetes, and nearly 90% of diabetic cases have Type 2 diabetes. The most commonly used diagnostic tests for prediabetes and Type 2 diabetes are the glycated hemoglobin (A1C) test, the fasting blood glucose (FBG) test, and the OGTT-all of which need patients to see a healthcare professional.
Yan Fossat, vice president of Klick Labs and the study’s principal investigator, stated that Klick’s non-intrusive and accessible technique can potentially test a huge number of people and help detect the high percentage of undiagnosed Type 2 diabetes patients.
Our research underscores the tremendous potential of voice technology in identifying Type 2 diabetes and other health conditions. Voice technology could revolutionize healthcare practices as an accessible and affordable digital screening tool.
Yan Fossat, Study Principal Investigator and Vice President, Klick
The next step, according to Fossat, will be to replicate the study and extend their research into other areas, such as prediabetes, women’s health, and hypertension, utilizing voice as a diagnosis.
This recent discovery is made possible by Klick Labs’ over a decade of knowledge and investment in machine learning, data science, and artificial intelligence across various therapeutic domains, including diabetes.
Their “Homeostasis as a proportional-integral control system” study, published in Nature Digital Medicine in 2020, used mathematical modeling to discover some of the fundamental differences in how glucose is regulated. Recently, their study “Screening for Impaired Glucose Homeostasis: A Novel Metric of Glycemic Control” was published in Mayo Clinic Proceedings: Digital Health.
Kaufman, J. M., et al. (2023) Screening for Impaired Glucose Homeostasis: A Novel Metric of Glycemic Control. Mayo Clinic Proceedings: Digital Health. doi:10.1016/j.mcpdig.2023.02.008
Veen, L. v., et al. (2023) Homeostasis as a proportional–integral control system. Nature Digital Medicine. doi:10.1038/s41746-020-0283-x