Could Your Smartphone Know You’re High?

Smartphone sensor data may be able to detect a cannabis high with up to a 90 percent accuracy, finds a new study by the Rutgers University Institute for Health, Health Care Policy and Aging Research.

The researchers found that travel and movement smartphone data were the most important information for detecting cannabis intoxication levels in a person, as reported in the advanced online November 2021 issue of Drug and Alcohol Dependence. 

Using smartphone features to detect cannabis high

Through self reports, phone data, and phone surveys, researchers could collect data on what their participants were doing and when, including how much cannabis they had consumed and how high they felt. 

This study looked at young adults, from 18 to 25 years old, who smoked or otherwise consumed weed at least twice a week. The research participants, who were all in Pittsburgh, Pennsylvania, had their data recorded daily for up to 30 days. This included self-reports of when they started and stopped cannabis use and how high they felt from 0 to 10 throughout the day. 

Our smartphones can play movies, create routes through traffic, send messages, among its many other uses, such as even taking a call once in a way. In this study, participants’ smartphones kept continuous sensory data on more than 100 features, including GPS, noise, light, and activity levels. 

And to really make sure that the participant’s data was fully captured and they couldn’t just chill, they also received phone surveys three times a day.

How fast do you move while high? 

The researchers created an AI to detect marijuana intoxication. Previously, the authors had developed machine-learning models to detect binge alcohol drinking using a smartphone app.

In this study, they found that the most important smartphone features to detect cannabis intoxication included travel, using GPS data to track where you’re going, and movement, the accelerometer data which tracks how fast you’re going. Likely, if you’re moving like a sloth between the bed, couch and the fridge, that travel and movement data could guess you’re high. 

In terms of limitations of this study, the major one is that only 57 young people were included. Though, interestingly enough, 58 percent of participants self-identified as female which is good in what can be a male-dominated research field.

The researchers note that current cannabis tests, using saliva, urine, or blood, have limitations in how well they measure the high and related impairment in daily life. This may be due to people that consume weed having widely different tolerance levels. A few puffs of a joint can give one person a slight buzz while making someone else feel very high. 

Now with medicated tinctures, drinks, and all matter of edibles, from mints to granola bars, it can be difficult to gauge dosages as well as effects and when they’ll hit. 

Using AI to predict harmful behaviour

Due to slowed response time in psychomotor functioning, getting high on weed can be dangerous in some situations, such as before driving or using heavy machinery. The researchers want this study to help towards developing a just-in-time adaptive intervention for cannabis users.

“Smartphones with mobile sensors are universal and can track our behavior in an unobtrusive way,” said lead author Sang Won Bae in a statement. “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.”