{{featured_button_text}}

The bees didn’t mind Colin Henderson’s microphone.

On Tuesday, the Missoula College professor plugged a long, narrow rod into his smartphone and slid it into one of the beehives he keeps at Fort Missoula. For 30 seconds, he set down his phone and let it record. An app he and his colleagues are developing, Bee Health Guru, analyzed the audio for signs of eight different bee colony problems. This hive was normal on all counts.

Getting that result through a physical inspection would have taken 15 minutes. Bee Health Guru did it in about a minute. For a beekeeper monitoring hundreds or thousands of hives for signs of disease, that “is a big time saver,” said Henderson.

He and other UM scientists, including Jerry Bromenshenk, David Firth and Robert Seccomb, have been studying bees’ biology for decades, training them to detect airborne toxins and even landmines, and forming a company, Bee Alert Technology Inc., to license their innovations. The Bee Health Guru app emerged from a years-long effort to gauge bee colonies’ health using one of their best-known traits: their sound.

A beehive’s familiar buzz, Henderson explained, is “mostly vibrations of the wings and thorax. Different diseases make them vibrate differently.”

In the mid-2000s, he said, “we started listening to colonies with different disorders. … We selected eight common sorts of infections and disorders in bee colonies that are bothering beekeepers.” They included the absence of a queen, the “foul brood” bacterial disease, and the increased aggression known as “Africanization.”

“We collected as much audio as we could” from colonies with these problems, Henderson explained, gathering reference material that could teach an artificial intelligence program to diagnose problems. The hardware wasn’t initially suited to the task: their first handheld recording devices, assembled in 2012, were “big, expensive, and clumsy to use,” the team wrote in a May story for Bee Culture magazine. Henderson estimated parts and labor for the initial model cost $1,000.

They soon turned to smartphones, launching an initial version of the Bee Health Guru app in fall 2017. Last year, increased processing power brought analysis time down to seconds.

“We’ve been in closed research mode,” Henderson said, with about 600 researchers and beekeepers worldwide using the app to monitor hives and collect data.

When the app completes its analysis, it displays the probability of each of the eight conditions, asks the user whether it agrees with the analysis, and uploads the data, along with GPS coordinates, to the cloud.

Right now, Henderson says, the app’s reached “70-85% successful discrimination of colony health.” They’re aiming for an 80% to 85% success rate, and go public by the end of the year.

That, he predicts, could be a boon to beekeepers, and the pollinators under their care. “If you know more about colonies, you can identify sick colonies and treat them more specifically, he said.

It likely won’t enable beekeepers to beat back colony collapse disorder, a phenomenon in which worker bees abandon the hive, leaving the queen and young to die. The percentage of hives affected has declined in recent years, but the Environmental Protection Agency still considers it a concern.

“Getting audio that you can fingerprint for that has been really hard,” Henderson said. “They collapse so quickly when they get the ailment that you’re usually looking after the fact.”

Nonetheless, their software has seen plenty of interest from the beekeeping world. In May, the team started a Kickstarter online fundraising campaign to get their product over the finish line. They set a $13,400 goal; as of Thursday, they had raised $28,286 from 653 backers.

“There are a lot of people that have a lot of interest in bee health,” Henderson said.

Get the latest local news delivered daily directly to your inbox!

* I understand and agree that registration on or use of this site constitutes agreement to its user agreement and privacy policy.
You must be logged in to react.
Click any reaction to login.
0
0
0
0
0