According to a new study, people could be diagnosed with diabetes using nothing more than a short voice recording from their phone.
By using an audio sample of only six to ten seconds, along with basic health information such as age, sex, height, and weight, scientists developed an AI model that can determine with nearly 90% accuracy whether someone has diabetes.
Klick Labs conducted the study, which involved 267 participants, including some who had already been diagnosed with type two diabetes.
Each participant was asked to record a phrase on their phone six times a day for two weeks, and the team used AI to analyze over 18,000 samples in order to identify acoustic differences between diabetics and non-diabetics.
These differences included pitch changes caused by type two diabetes, which cannot be detected by the human ear.
The model achieved an accuracy rate of 89% for women and 86% for men.
Study author Jaycee Kaufman stated that the results could “revolutionize” diabetes screening methods.
In the UK, over 90% of adults with diabetes have type two, but many remain unaware of their condition for years due to the general or non-existent symptoms.
Currently, individuals need to visit a GP and undergo urine and blood tests to be tested for diabetes.
Ms. Kaufman said, “Current detection methods often require a significant amount of time, travel, and expense.
“Voice technology has the potential to eliminate these barriers entirely.”
Prior research has demonstrated that voice recordings, combined with AI, can be used to diagnose other illnesses, including COVID-19.
Klick Labs believes that this technology could also be used to diagnose conditions such as prediabetes and hypertension.
The peer-reviewed study has been published in the Mayo Clinic Proceedings journal.