New research has found that diabetes may be diagnosed with just a short audio recording from a mobile phone.
Scientists can determine whether someone has diabetes with nearly 90% accuracy using just a 6-10 second audio sample and basic health data such as age, gender, height, and weight I created an AI model.
Klick Labs recruited 267 people for the study, including some who had already been diagnosed with type 2 diabetes.
Each subject was asked to record a phrase on their phone six times a day for two weeks, and the team used AI to analyze more than 18,000 samples to determine the acoustic differences between diabetics and non-diabetics. I looked into it.
These included changes in pitch caused by type 2 diabetes that are imperceptible to the human ear.
This model had an accuracy rate of 89% for women and 86% for men.
Study author Jaycee Kaufman said the results could “change” the way we screen for diabetes.
More than 90% of adults with diabetes in the UK have type 2 diabetes, but many go undetected for years as symptoms may be systemic or absent. I am.
Testing for this disease usually requires a visit to a general practitioner and urine and blood tests.
“Current detection methods can be time-consuming, travel-intensive, and costly,” Kaufman said.
“Voice technology has the potential to completely remove these barriers.”
Read more science and technology news:
The rise and fall of BlackBerry
Debunking myths about the influenza vaccine
NASA reveals ‘unbelievable’ asteroid discovery
Previous research has shown that audio recordings can be used in conjunction with AI to diagnose other diseases. Including new coronavirus infection.
Klick Labs believes this technology can also diagnose conditions such as prediabetes and hypertension.
This peer-reviewed study was published in the Mayo Clinic Proceedings journal.