Mini radar and AI instead of finger-pricking for diabetics?


“We want to sense blood inside the body without actually having to sample any fluid,” said engineer Professor George Shaker. “Our hope is this can be realised as a smart watch to monitor glucose continuously.”

The proof-of-concept involved re-purposing a miniature 60GHz gesture-recognition radar created by Google and Infineon – Project Soli – both firms are participating in the Waterloo research – to investigate the dielectric properties of samples which mimic blood glucose concentrations in diabetic patients.

It involves bouncing radio waves within the samples, and then using artificial intelligence to un-pick data from the returns.

“The software is capable of detecting glucose changes based on more than 500 wave features or characteristics, including how long it takes for them to bounce back to the device,” said the University.

In tests outside humans, using different concentrations of glucose in substances and structures that mimic human tissue, results were 85% as accurate as traditional blood analysis. “The correlation was actually amazing,” said Shaker.

Results are published in ‘Non-invasive monitoring of glucose level changes utilizing a mm-wave radar system‘ in the International Journal of Mobile Human Computer Interaction (IJMHCI).

Further research is multi-pronged, intending to:

  • improve accuracy
  • operate through skin
  • with Infineon, shrink the radar and make it cheaper
  • run the AI algorithms inside a watch, rather than on a remote computer

“I’m hoping we’ll see a wearable device on the market within the next five years,” said Shaker. “There are challenges, but the research has been going at a really good rate.”

Completing non-blood-sample technologies include: through-skin IR or near-IR absorbance spectroscopy, and analysis of breath, saliva, sweat or tears.