• MIT researchers built a shoebox sized device that reads glucose without breaking the skin
  • The system uses Raman spectroscopy and a band pass approach to pick out glucose signals
  • In a pilot test, readings closely tracked two standard continuous glucose monitors

Finger pricks and sensor insertions have long been part of life with diabetes.

Now, a team at the Massachusetts Institute of Technology has taken a significant step towards true noninvasive glucose monitoring.

Their compact device uses light to read glucose levels through the skin in about half a minute.

The study, published in Analytical Chemistry, describes a new way to apply Raman spectroscopy so that weak glucose signals can be measured reliably without complex artificial intelligence.

Why noninvasive monitoring is so difficult

For people with diabetes, frequent glucose checks are essential to avoid both high and low levels and to reduce long term complications.

Classic finger prick tests are accurate but painful and impractical for near continuous use.

Existing continuous glucose monitors rely on sensors inserted into the skin and need regular replacement.

Many groups have tried to read glucose using light, breath, or other indirect markers.

So far, none of these methods has produced a noninvasive device that is accurate enough, robust enough and affordable enough for everyday clinical use.

Raman spectroscopy, which analyses how light scatters off molecules, is attractive because it can in principle identify glucose directly.

In practice, the glucose signal is tiny compared with other signals from skin and tissue, and it is easily lost in noise.

A focused band pass Raman approach

The MIT team designed a system that shines near infrared light onto the skin and collects the returning signal at a slightly different angle.

This geometry helps reduce unwanted scattered light and makes glucose related features easier to detect.

Instead of recording the entire Raman spectrum, which would involve around a thousand bands, the researchers concentrated on three carefully chosen bands around a known glucose peak.

Two side bands were used as an internal reference to help account for background changes.

By focusing on this narrow region, they could simplify the optics and electronics and avoid the need for heavy machine learning models.

The result is a device about the size of a shoebox that sits on a table. The user rests an arm on the top and a beam of light passes through a small window to the skin. Each reading takes around 36 seconds.

Testing the prototype

To test the system, the researchers first used artificial tissue samples with controlled glucose concentrations.

They then moved to a pilot study in a twenty seven year old healthy volunteer.

During the human test, the participant consumed two glucose drinks to create rises and falls in blood sugar.

The Raman device took readings every five minutes over four hours.

For comparison, two commercial invasive continuous glucose monitors were inserted into the other arm and standard finger prick tests were carried out every ten minutes.

The noninvasive device tracked changes in glucose closely and its results aligned well with those from the established devices. In this single participant, performance metrics were within the range seen with commercial continuous monitors.

Get our free newsletters

Stay up to date with the latest news, research and breakthroughs.