Imagine testing your blood glucose by taking a photo of your eye. No fingersticks, test strips, or calibrating required. That’s the idea behind EasyGlucose, Bryan Chiang’s winning entry in the 2019 Microsoft Imagine Cup, a global contest for student developers.
Chiang, an 18-year-old freshman at the University of California, Los Angeles, said he conceived the idea for the cloud-based app after watching his grandmother struggle with managing her diabetes. According to information from Microsoft, app users capture high-resolution images of their eye with a smartphone through a custom, low-cost iris imaging adapter, which then feeds the image to the app for analysis.
“Our blood glucose levels are actually highly correlated with the glucose levels in the eye,” Chiang said during the competition Monday, according to a statement
. “And by analyzing images of the eye, we can determine our glucose levels by looking at specific structures inside the iris.”
The system uses a deep learning computer vision framework developed with Azure Virtual Machines, which analyzes variation in the iris from the image to predict a person’s blood glucose level. Chiang’s app has achieved an error rate of 6.93%, surpassing existing noninvasive methods.
EasyGlucose is comparatively cheap: it has a one-time $10 cost for the lens adapter that connects to the user’s smartphone, along with a $20 per month subscription fee. While newer glucose monitoring methods on the market have largely eliminated the need for fingersticks, they cost more and may require a sensor to be worn on the arm.
The young developer, who won $100,000 in the competition, is wasting no time bringing his idea to the market: he has a patent pending for app’s framework and will do more tests with the product at Stanford University before filing for approval with FDA.
“Ultimately, by reducing the pain and cost of current methods, EasyGlucose provides a comfortable, user-friendly, non-invasive and accurate way for diabetic patients to manage their blood sugar levels at a fraction of the cost of today's methods,” Microsoft said in a statement