A program uses available data and ten seconds of speech to identify if someone has diabetes. Moreover, the study has found that it is 7 out of 8 times accurate. However, it should be available on smartphones soon to provide an accessible option for people with limited access to healthcare.
The team at Klick Labs made 267 people record a short phrase on their phone six times a day for fourteen days. All of them had undergone standard Type 2 diabetes testing. Moreover, the researchers then looked for acoustic differences between those who tested positive and negative.
The researchers combined the absence and presence of identified features in voice prints. In addition to the participant’s age, gender, height, and weight. The artificial intelligence model predicted each participant’s status. It was 86% accurate for men and 89% accurate for women.
The researchers wrote,
Voice synthesis is a complex process that relies on the combined effects of the respiratory system, the nervous system, and the larynx. Anything that affects these systems can influence the voice,
They also wrote that most people will be unable to identify these changes. However, computers can perform the required analysis.
The predictive tool that was most powerful was the variation in pitch between the times the phrase was recorded. Furthermore, the other method that added to the accuracy was “perturbation jitter,” but for only one sex. It was predictive for women and amplitude “perturbation quotient shimmer” for men.
The first author of the study, Jaycee Kaufman, said,
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.
The co-author of the study, Yan Fossat said,
Voice technology could revolutionize healthcare practices as an accessible and affordable digital screening tool,
