
Three trained bees in a special harness that holds them in place for the diabetes-detecting experiment. (Photo courtesy of Juliet Phillips, Bee Healthy project.)
By Richard Knox
The latest book by humorist David Sedaris is implausibly titled “Let’s Explore Diabetes with Owls.” But as we all know, life is stranger than literature: Now, an imaginative team of social entrepreneurs has devised a way to explore diabetes with bees — that is, to train honeybees to diagnose hidden cases of diabetes.
It’s no crackpot idea. It’s in the running with five other finalists for a million-dollar prize given each year by the Clinton Foundation.
Here’s the concept: Bees are 10,000 times more sensitive to chemicals in the air than humans. The breath of humans with diabetes contains higher levels of a chemical called acetone. Bees are easily trained to stick out their tongues when they detect a certain concentration of acetone.
Put a bunch of these trained bees into tiny harnesses, have a person breathe into a straw aimed at the constrained bees and voila! A diabetes screening system that doesn’t require laboratories, expensive machines, highly trained technicians, dietary fasting, or more than a modicum of money.
It may be a good way to screen large numbers of people for undiagnosed diabetes in developing countries such as India, where the disease is burgeoning even faster than in overfed America.

Juliet Phillips of the Bee Healthy project studies a bee in harness, ready to be bathed in the breath of a volunteer. The bee is trained to stick out its tongue if it scents a certain level of a chemical in the volunteer’s breath, signaling diabetes. (Photo courtesy of Juliet Phillips, Bee Healthy project)
“Millions of people aren’t aware they have this disease,” says Juliet Phillips, a leader of the project, called “Bee Healthy.” “They aren’t even aware there is this disease. So there’s a need to screen people for diabetes that’s free for people in slums but also culturally acceptable.”
Phillips and her colleague Tobias Horstmann were in Boston this month to test the idea on a group of people with known diabetes. The experiment, at the Joslin Diabetes Center, found that bees could identify the diabetic patients 70 percent of the time.
“That’s not as high as we want to go, but we believe we can get there,” Horstmann says. “We can get improvement in the training of bees.”
In addition, the Boston patients in the test all had well-controlled diabetes, so the level of acetone in their breath was much lower than undetected diabetics in a developing country whose diabetes is out of control. Continue reading