A new study used artificial intelligence to predict cannabis disorder with surprisingly high accuracy based on biometric data captured by smartphone sensors.
Researchers at the Stevens Institute of Technology recently published the following paper. study in drug and alcohol addiction We analyzed smartphone data from cannabis users and non-cannabis users. Cannabis users self-reported the amount of time they consumed the drug and the level of intoxication they experienced based on a simple 1-10 scale.
By comparing and contrasting over 100 different sensory inputs, including time, location, noise, and movement levels, picked up from the cell phones of cannabis users and non-cannabis users, the researchers found that They claim to have identified significant differences between datasets among users. The marijuana user was intoxicated and there was a difference that normal human senses could not discern. The same technology is also used to study and predict alcohol and other drug-induced disorders.
“Smartphones with mobile sensors are versatile and can track our movements in a discreet way,” said Sang Won Bae, an assistant professor at Stevens Institute of Technology, who led the study. “It’s not distracting, you don’t have to wear it, and the data it collects could help prevent poor decision-making when under the influence.”
Differences in the datasets are used to train artificial intelligence learning models that may one day be able to detect whether someone is under the influence of cannabis in real time through information detected by cell phone sensors. not. This could hypothetically allow the phone to intervene in some way, in the form of notifications suggesting things like ride-sharing services. Researchers claimed that an AI model could predict cannabis addiction with 90% accuracy after being trained on smartphone data.
“It’s important to give people a chance to change their behavior before something negative happens,” Bae said. “This research aims to predict human behavior as a way to support people with physical or cognitive disabilities.”
This study claimed that cannabis disorders could be predicted with approximately 67% accuracy using only smartphone data, but when combined with time data such as day of the week and time of day, artificial intelligence learning model “Light Gradient Boosting Machine” can now predict cannabis disorders with significantly improved accuracy of 90%. Disability was measured using a 0–10 scale. A score of 0 means “not intoxicated,” a score of 1-3 is considered “low intoxication,” and a score of 4-10 means “moderate” level of intoxication.
“Because temporal features alone have the potential to predict the ‘routine’ of cannabis intoxication, we tested the importance of temporal features (i.e., day of week, time of day) compared to smartphone sensor data using model performance alone.” says the study. AI learning models can predict failures with 60% accuracy based solely on the time factor.
There are several notable limitations to this data collection method that may have influenced the results, including small population size and reporting bias, among other factors. The study monitored smartphone data from 57 cannabis users who consumed cannabis on a total of 451 different occasions throughout the study. Duration of ingestion and level of intoxication were also self-reported by participants, which is, of course, a highly subjective experience for each user. Both of these factors may have some influence on the study results, which the study authors acknowledged.
This is not the first attempt to detect cannabis-induced impairment in real time. Most blood, saliva, and urine tests cannot predict current impairment, only those that have been used recently. A Montana-based company has been working on releasing an eye scanner for police use in the field that detects marijuana impairment by analyzing eye movements, but it has not yet been made available to the public. In any case, there are very few, if any, options for accurately detecting the level of impairment or the timing of impairment, and research claims that it at least proves achievable. Masu.
“This exploratory study demonstrated the feasibility of using smartphone sensor data to detect subjective cannabis intoxication in young people in natural settings,” the study states. “Smartphone sensor data contributed unique information and long-lasting ability to detect subjective cannabis intoxication.”
Cannabis users don’t need to hide their phones before getting stoned just yet, as the results of this study are still preliminary and require further study before a definitive assessment can be made.