Personalized sweat sensor reliably monitors blood glucose without finger pricks

[Northerner]

Many people with diabetes endure multiple, painful finger pricks each day to measure their blood glucose. Now, researchers reporting in ACS Sensors have developed a device that can measure glucose in sweat with the touch of a fingertip, and then a personalized algorithm provides an accurate estimate of blood glucose levels.

According to the American Diabetes Association, more than 34 million children and adults in the U.S. have diabetes. Although self-monitoring of blood glucose is a critical part of diabetes management, the pain and inconvenience caused by finger-stick blood sampling can keep people from testing as often as they should. Scientists have developed ways to measure glucose in sweat, but because levels of the sugar are much lower than in blood, they can vary with a person's sweat rate and skin properties. As a result, the glucose level in sweat usually doesn't accurately reflect the value in blood. To obtain a more reliable estimate of blood sugar from sweat, Joseph Wang and colleagues wanted to devise a system that could collect sweat from a fingertip, measure glucose and then correct for individual variability.

Personalized sweat sensor reliably monitors blood glucose without finger pricks

Many people with diabetes endure multiple, painful finger pricks each day to measure their blood glucose. Now, researchers have developed a device that can measure glucose in sweat with the touch of a fingertip, and then a personalized algorithm provides an accurate estimate of blood glucose levels.

www.sciencedaily.com

 

[mikeyB]

Don't think the DVLA will be accepting a sweaty finger BG test any time soon as acceptable before driving. And how do they account for individual variability? Calibrating it with fingertip blood tests? And to stop contamination with other stuff your fingers get into, what do you do? Wash your hands thoroughly, removing all the sweat.

I bet the system can't get a CE mark.

 

[Bruce Stephens]

And how do they account for individual variability? Calibrating it with fingertip blood tests? And to stop contamination with other stuff your fingers get into, what do you do? Wash your hands thoroughly, removing all the sweat.

It requires an individually calibrated algorithm of some sort, yes.

The article says "In tests, the algorithm was more than 95% accurate in predicting blood glucose levels before and after meals. To calibrate the device, a person with diabetes would need a finger prick only once or twice per month."

But yes, it feels like another case of hoping that if you throw enough relevant (but rather fuzzy) data at a machine learning algorithm you can get an accurate result. I won't hold my breath for this to end up in a useful product.

(I'm still seeing occasional news stories saying that the next version of the Apple Watch will have glucose monitoring.)