New AI can detect if humans have smoked cannabis and ‘alert medical professionals’ by observing ‘micromovements’
A NEW AI model may be able to detect if someone has recently used cannabis.
The advancement could one day help to assist medical professionals in quickly determining if someone in need of medical care has taken cannabis.
Currently, professionals use urine, saliva, or a hair strand to test for cannabis use.
But these lab results can take a few days to analyze and come back.
This prompted researcher Sang Won Bae from the Stevens Institute of Technology and a team of colleagues to look for a way to quickly determine whether someone may be on cannabis.
For their research, they looked at 33 adults who used cannabis at least twice a week.
They had the participants report their use every day for up to 30 days as well as wear an activity tracker.
This tracker collected information like their heart rates, step counts and sleep quality.
Other sensors on the participants’ phones provided data on their micromovements.
Specifically, this meant the way they held their phone, which helped determined their stability and coordination.
Data collected from some of these participants was then fed to an AI to determine if someone may have used cannabis.
Researchers subsequently used the AI on the remaining participants’ data to test its capability.
Their results showed that the AI was 85 percent accurate at detecting someone who had been moderately high within the past 5 minutes.
However, some experts remain skeptical at the results and believe more research needs to be conducted.
Joseph Wu at Stanford University in California and Mark Chandy at Western University in Canada both agree that a larger group of subjects needs to be tested, per New Scientist.
Moreover, some experts expressed concerns about users reliability on reporting their cannabis use.