HOBOKEN, N.J. — Smartphone apps regularly track what their users are doing and when they do it. Now, researchers from the Stevens Institute of Technology have developed a smartphone sensor that can tell if you’re high on marijuana.
Assistant professor Sang Won Bae previously created machine learning technology that can detect if someone has been binge drinking. The new AI system detects marijuana intoxication by examining smartphone user behavior before and after using cannabis.
“Smartphones with mobile sensors are universal and can track our behavior in an unobtrusive way,” Bae says in a university release. “They are not a distraction, you don’t have to wear them, and the data they collect can potentially prevent poor decision-making when under the influence.”
Finding who’s high in real time
Study authors explain that traditional drug testing methods — which examine blood, urine, or saliva — have limitations and are generally slow to process. Since marijuana can impair psychomotor functioning when users are experiencing that classic “high,” the team says their study reveals how technology can help provide real time intervention for users who may be putting themselves at risk of injury.
Study authors tracked smartphone motion sensors in the devices of volunteers who reported using cannabis at least twice a week. The team used over 100 features to track whether each participant was high. Those features included GPS, noise, light, and activity trackers. The team then looked at smartphone data usage among these individuals at the times they reported they were either “high” or “sober.”
Bae and her colleagues from Rutgers and Carnegie Mellon University discovered the sensor could spot behavioral differences that predicted marijuana intoxication with up to 90 percent accuracy. The Stevens researcher believes AI software that detects marijuana intoxication will protect cannabis users from making risky decisions while their reflexes are impaired.
“It’s important to give people the chance to change their behavior before something negative happens,” Bae concludes. “This study aims to predict human behavior as a way to support people while physically or cognitively impaired.”
The study appears in the journal Drug and Alcohol Dependence.