This Pittsburgh startup wants to use AI to stop opioid relapse before it happens

Behaivior is one of 147 teams in the IBM Watson AI XPRIZE contest.

Three-Fourths of the Behaivior team: Jeremy Guttman, Ellie Gordon and Ryan O’Shea

Three-Fourths of the Behaivior team: Jeremy Guttman, Ellie Gordon and Ryan O’Shea

Courtesy of Behaivior
MJ Slaby

A new Pittsburgh startup wants to know when recovering opioid users are going to relapse before they do.

So the founders of Behaivior are using artificial intelligence and a wearable device to do just that. The device would gather data and, using a predictive model, find and gauge the likelihood of a drug relapse for the person wearing it. If a relapse is possible, a message or alert would be sent for human intervention.

For example: Maybe stress is a trigger for drug use, so the device could sense when the wearer’s heart rate goes up or when they start to sweat, said Ryan O’Shea, one of Behaivior’s founders. Or if it’s linked to the person’s phone, the device would be able to tell that a person missed a counseling appointment — using the phone’s calendar and GPS — and might be at risk of a relapse, he said.

Instead of a parol officer or drug counselor finding out about a relapse after it happens, the device would send them an alert and allow them to intervene, O’Shea said. The device could help keep people out of court for relapses and related crimes, and it’s another alternative to expensive and often full treatment centers.

On Tuesday, Behaivior was one of 147 worldwide teams accepted into the IBM Watson AI XPRIZE, a $5 million contest focused on finding ways for humans to use artificial intelligence to solve big problems. The competition lasts four years with “milestone competitions in 2017 and 2018,” followed by the top three teams competing for the top prize at TED 2020, per the XPRIZE website. This gives the team access to mentors at IBM and a variety of resources, O’Shea said.

But it’s still all new. Behaivior was founded about four months ago. (Yes, there’s an extra “i”  — representing AI in the name — but it’s pronounced “behavior”.)

O’Shea and fellow founders Jeremy Guttman, Ellie Gordon and James Tiu met at an AI hackathon in January at Ascender and started working together. There, they earned having their XPRIZE application fee paid.

The group took two ideas — wanting to help with treatment for opioid use and using machine learning to predict behavior — and brought them together.

“Nothing about what we’re doing is a novel idea,” O’Shea said, adding that insurance companies and social media companies already try to predict behavior, but AI is the future.

The team also won to the first ever Social Justice Innovation Weekend in February, where they competed as RecoverEQ. There, they met Courtney McFeaters, who is three years sober from a heroin addiction, and she joined the team for the weekend. Both McFeaters and O’Shea said they plan to continue to work together as the project advances.

Right now, they’re working to decide if they want to adapt a wearable device that exists or create their own, O’Shea said, adding that the company still has to collect data and create its algorithm. The goal is to expand once they have a product. O’Shea said the device is creating “general pattern recognition artificial intelligence” that can be used for multiple habits.

“Ultimately this has the potential to help reduce stress, smoking, overeating, sleeplessness, and more. The possibilities are endless,” he said.