Diabetes

Could radar put an end to diabetic finger-pricking?

Thanks to a radar-based system that's currently in development, doing THIS may someday no longer be necessary
Thanks to a radar-based system that's currently in development, doing THIS may someday no longer be necessary

Although finger-prick blood glucose tests are a daily necessity for millions of diabetics, a less-painful alternative may be on the horizon. Led by Prof. George Shaker, a team from Canada's University of Waterloo is looking at using radar and artificial intelligence (AI) to do the job.

The project started with Google and German company Infineon jointly creating a small radar device known as Soli, and then inviting teams from around the world to suggest possible applications for it. Shaker's group responded by incorporating the device into a system that non-invasively measures glucose levels within various liquids.

It does so by sending high-frequency radio waves into those liquids, and then using AI algorithms to analyze the manner in which those waves are reflected – over 500 wave features are scrutinized, including the amount of time that elapses between the waves being emitted by the Soli device and then bouncing back to it.

In lab tests in which the system was used to determine the blood glucose levels of human volunteers, it proved to be 85 percent as accurate as traditional finger-prick testing.

It is hoped that its accuracy can be significantly boosted, once the technical challenge of making readings through the skin is overcome. The scientists are also working with Infineon on miniaturizing the technology, along with reducing its cost and power requirements, to the point that it could one day conceivably be incorporated into something such as a smartwatch.

"I'm hoping we'll see a wearable device on the market within the next five years," says Shaker. "There are challenges, but the research has been going at a really good rate." A paper on that research was recently published in the International Journal of Mobile Human-Computer Interaction.

Various other recently-developed alternatives to finger-pricking include experimental systems that utilize lasers, graphene-based skin patches, and electrified temporary tattoos.

Source: University of Waterloo

  • Facebook
  • Twitter
  • Flipboard
  • LinkedIn
1 comment
christopher
Put the dataset on Kaggle - that will fix everything !