Researchers entered data from both methods into a computer program and looked for patterns within the electromagnetic data that corresponded to the glucose readings from the blood samples.

Although the early results are promising, it remains to be seen how broadly they can be generalized. "We're still working on verifying that the calibrations are truly robust," says Jean. "In other words, the data looks good for the people we've had a lot of experience with, but now we have to make sure that if a new thumb comes along, it works on that one."

What's more, Jean's sample pool tended to exhibit glucose levels within the normal range. To verify the sensor's accuracy, the team needs to test it on volunteers with varying glucose levels. In the next few months, Jean plans to test the sensor on patients at Scott and White Hospital, in Temple, TX, whose glucose readings may be "all over the map."

"If [a monitor] could be developed, it would be enormously promising because it's not just noninvasive but could give continuous data," says Howard Wolpert, director of the Insulin Pump Program at the Joslin Diabetes Center, which is based in Boston. "That's what people are interested in, because with devices today, you're only looking at intermittent time points, and glucose fluctuations can be quite dramatic."

Jean says that while his ultimate goal is to design an accurate sensor cheap enough for patients to carry around with them, he expects that one of the first early uses of the technology will be as screening devices at local drugstores, much like the large commercial monitors that take blood-pressure and heart-rate readings.

"It could provide a useful service for someone who didn't know they're diabetic, and you could say, 'Your blood sugar is kind of high. You should go to the doctor,'" says Jean.