Wearable weapons in drug war
Local startup and others developing variety of Fitbit-like devices in an effort to prevent opioid overdoses
A wearable just like your Fitbit could be the solution to America’s opioid epidemic. Or at least that’s what a Lawrenceville-based startup is betting on.
Instead of logging your steps, stairs and calories, Behaivior — spelled with “AI” in the middle for its artificial intelligence software — wants to track your location, heart rate and stress levels, among other variables. The idea is to see if people addicted to opioids such as heroin are at risk of relapsing.
It all began at Ascender, an East Liberty-based co-working space, when a defense attorney working in West Virgina pitched the idea for a wearable to help prevent overdoses.
“Many of his clients were struggling with heroin addiction. And just like in Pittsburgh, West Virginia and Ohio are hit really hard by the opioid epidemic,” said Ellie Gordon, founder and CEO of Behaivior.
That attorney is no longer with the company, but the remaining five full-time employees are competing in the IBM Watson AI XPrize, a global competition to develop artificial intelligence-based solutions to realworld problems. There’s a $5 million purse up for grabs.
While Behaivior is the only startup working on such a solution in the XPrize competition, it’s joined by a brigade of other companies working on wearable solutions for those recovering from opioid abuse — each with a slightly different approach.
A mountain of competitors
According to figures for 2017 from the U.S. Department of Health and Human Services, an estimated 2.1 million people had an opioid use disorder and 47,600 people died from overdose.
Just as the opioid epidemic permeates every region of the country, researchers from all over the U.S. are tapping into ways wearable technology can help solve this public health crisis.
In Waltham, Mass., digital medicine company NeuroMetrix has built a wearable device called Quell. Drug-free and worn on the calf, the device sends neural pulses to the brain, releasing pain blockers called enkephalins. It alleviates pain while serving as a stand-in for opioid medications.
DyAnsys Inc., based in San Mateo, Calif., created “Drug Relief,” an aid to ease withdrawal without the use of other opioids. It sends electrical pulses to the ear through tiny needles to alleviate anxiety, depression, nausea and cravings.
Researchers at Penn State University have started Wear-IT, a startup company fusing data from existing wearables and a user’s cell phone to determine triggers that could cause relapse, very similar to Behaivior’s own model. If an addict’s heart rate increases near a drug store, for example, Wear-IT may suggest an alternate commute home from work.
Not far from Behaivior’s office, a group of Carnegie Mellon University graduate students came up with a similar device.
The “HopeBand” uses pulse oximetry, a noninvasive test to determine oxygen saturation levels in the blood. Sensors on the band take repeated readings of the wearer’s blood oxygen level when paired to a smartphone via a wireless connection.
If a person’s blood oxygen level drops for more than 30 seconds, an LED light changes from green to red. Through the connected smartphone, the HopeBand can send a message — complete with the wearer’s GPS location — to emergency contacts.
Picking up the patterns
By contrast, Behaivior will measure multiple indicators through sensors — including movement, location and galvanic skin response (a stress measure), sleep and heart rate.
The idea is to reduce false positives that could trigger an unnecessary emergency response.
Your heart rate can go up for myriad reasons, Ms. Gordon pointed out. Through movement measurements, Behaivior can tell that someone is exercising, not craving drugs.
“When you want to achieve high accuracy levels, you want to have more data coming in, you can have higher accuracy levels,” she explained. “Everyone’s bodies are different. So for some things, peoples’ responses won’t be as predictive.”
What differentiates Behaivior from its competitors, Ms. Gordon said, is the artificial intelligencedriven pattern recognition that helps sort through the information.
Through machine learning, Behaivior’s software platform can fuse together seemingly unrelated events.
If someone who consistently attends support group meetings every week abruptly stops going, and if the system has a geolocation for that meeting, that’s telling, she said. And if that person is also staying up really late and smoking cigarettes, that’s another red flag.
“If somebody sat down with all of those data points, they may be able to determine that themselves. But they probably couldn’t,” Ms. Gordon said. “It’s doing what doctors don’t have the time or the ability to do.”
Journey to the shelf
In the competitive medical devices space, it’s beneficial for consumers to see multiple versions of similar concepts, said Michael Hufford, co-founder and CEO of Harm Reduction Therapeutics, a Squirrel Hill-based nonprofit pharmaceuticals company.
“The market is very good at shaking those things out,” he said. “Whoever is trying to tap this market needs to make it at the lowest possible price.”
Mr. Hufford, whose firm sponsored the CMU students working on the HopeBand, said that because the U.S. Food and Drug Administration usually must approve medical devices in a lengthy process and because companies also must secure mass manufacturing contracts, it’s tough to bring even the most brilliant wearable ideas to hospitals.
He’s particularly cautious when it comes to conveying that the HopeBand will soon come to a store near you.
“We get lots of emails from family members [of addicts],” he said. “I’m personally very cognizant of not wanting to raise hopes about how soon these things may be available.”
Ms. Gordon said the Behaivior wearable will not need FDA approval and could be available as early as this summer.
“It’s doing what doctors don’t have the time or the ability to do.” — Ellie Gordon, speaking about Behaivior’s software platform
Since Behaivior could ultimately classify itself as a “general wellness” system, rather than a medical device, per FDA guidelines — it’s not implanted, invasive, nor runs the risk of causing harm — Ms. Gordon said it does not need to go through the lengthy and expensive approval process.
It’s a fine line, though. Last fall, the fourth iteration of the Apple Watch became the first FDA-approved smart watch. With high confidence, it can detect a heart condition called atrial fibrillation, or AFib.
And in 2017, the FDA unveiled a “fast track” program for technology companies like Apple, Samsung and Fitbit to encourage health care innovation.
Behaivior — which has received funding from the National Institutes of Health, National Science Foundation, CMU and East Liberty-based accelerator AlphaLab — will be device agnostic. That means no lengthy timeline to develop hardware, Ms. Gordon said; the company could theoretically use its software on any device with the right sensor capabilities.
She envisions selling the device to rehab centers for about $50 per month, a recurring fee to use the software. Ms. Gordon hopes to see insurance companies subsidizing the device and says even municipal government may have a role. Cities are already paying for needle exchanges and could add this device to their offerings.
Still, Behaivior must wade through a sea of competitors.
Philip Brooks, an entrepreneur-in-residence at the University of Pittsburgh’s Innovation Institute, said that’s not necessarily a downer for the startup.
He said that because approaches to wearable devices for the addiction recovery space are varied and people with substance abuse issues may react to those interventions differently, there may be room for multiple “winners” to land a spot on the shelf.
If a company is focused on prevention, for example, some people with substance abuse issues may positively respond if the device triggers a message from their support group. Others might ignore the message and use again.
“The market is so big and the problem so massive that I think there are opportunities,” Mr. Brooks said. “Frankly, there are probably multiple solutions that are out there.”