Fact Check: How Can Artificial Intelligence Detect Diabetes?

New Delhi: Artificial intelligence (AI) is making such rapid strikes that it is almost scary.

It’s being applied in all fields, including in medical sciences.

According to a new study, AI can analyse speech patterns to detect type 2 diabetes with great accuracy.

The report, published in the Mayo Clinic Proceedings: Digital Health medical journal, a short voice recording can determine whether an individual has diabetes.

The recently-developed AI screens voice recordings of 6 to 10 seconds, and looks for differences in vocal pitch and intensity. The programme combines with basic data like age, gender, height and weight to gauge whether the speaker has type 2 diabetes.

Because of vocal variances between male and female speakers, the tests were 89% accurate in examining females and 86% for males.

To develop the AI programme, Jaycee Kaufman and her team at Canada’s Ontario Tech University recorded voices of 267 persons who did not have diabetes or had already been diagnosed with type 2 diabetes.

Participants recorded a short sentence six times every day on their smartphone over a fortnight. Around 18,000 voice samples were generated, from which 14 acoustic features were identified as they differed between participants with and without diabetes.

“Current methods of detection can require a lot of time, travel and cost. Voice technology has the potential to remove these barriers entirely,” said Kaufmann, a research scientist at Klick Labs.

The AI method, which could prove to be a useful diagnostic tool, is expected to help identify humans with undiagnosed diabetes.

There are around 240 million adults around the world who are not aware that they have diabetes.

According to the International Diabetes Federation, nearly 90% of cases are type 2 diabetes, which have an elevated risk of cardiovascular diseases like heart attack, stroke and poor circulation in the legs and feet.

Get real time updates directly on you device, subscribe now.

Comments are closed.