I was surfing the web to find a good project to make, using Deep Learning applied to the Medical Sciences and I saw a freshman that built an app that can reads the glucose levels only with a picture of the eye! See: https://www.youtube.com/watch?v=tl3fld7cxGA&t=1119s

I was intrigued by it and starting to search for scientific articles to find which part of the eye is responsible for "affecting" glucose levels' readings. To my surprise, I haven't found anything! Only a good application of neural networks to identify diabetic retinopathy, which makes sense by analysing pictures of the eye.

Does anyone have any knowledge to explain me how I can measure (or with what should my measures be based with) glucose levels with the eye? Or have any articles that provide such information?


  • 1
    The video explains it and also says that Bryan Chiang has a patent pending for it. Jul 30 '20 at 22:21
  • 1
    @BryanKrause Yes, I understand but it's not clear. Did you really understand what was he analysing on the eye? The patent part is bluff as there is no patent under his name neither the glucose project (from what I've searched for). Jul 30 '20 at 22:42
  • 2
    @TomásBettencourt I doubt the patent claim is a bluff. He seems to have Stanford backing, so the patent is probably filed under the collaborating researcher's name. I doubt you're going to find implementation details publicly available until the patent is complete.
    – Carey Gregory
    Jul 31 '20 at 1:06
  • 2
    The eye doesn't affect glucose levels, but glucose levels affect the eye. My guess is that since intraocular pressure varies with serum glucose (and rapidly so), this is what the device will be measuring by comparing crypt depth, etc. It would be quite the feat. Jul 31 '20 at 6:06
  • 2
    I’m voting to close this question because the answer is provided in the sources provided in the question.
    – Thomas
    Jul 31 '20 at 6:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.