The functions of smart phones are becoming more and more comprehensive, and it is almost necessary to take care of daily life, and of course also our health. Some people use it to record daily exercise status, while others use apps to monitor their heart and lung function.
But this is not enough. Scientists can always think of even more bizarre needs: in the future, your mobile phone will also have the opportunity to become a “marijuana detector”! The research team of the Rutgers Institute for Health, Health Care Policy and Aging Research investigates the “highness” of cannabis users and combines it with machine learning technology to create A daily gadget that can accurately determine the degree of cannabis poisoning.
The study was published in “Drug and Alcohol Dependence” (Drug and Alcohol Dependence) Periodical[1]。
French comedy series “Cannabis Cafe” (Family Business), young entrepreneur Joseph. Joseph Hazan (Joseph Hazan) took a ride on the legalization of marijuana in France and decided to take his old father Gerhard. Gérard Hazan’s butcher shop was transformed into a cannabis cafe. Although my father was very opposed at first, but in the idol Enrico.Enrico Macias (Enrico Macias)Follow the wayAt last, his father also experienced the pleasure of Houma, and began to actively face their café business.
Why cannabis can make peopleHappy to give up persistence? A guy called Tetrahydrocannabinol (THC) plays a key role.Because the chemical structure of THC is very similar to the human body’s endogenous cannabis “anandamide”, it can bind to cannabinoid receptors and activate the brain’s reward system, making us feel physically and mentally happy[2]。
This inevitably makes people curious: how do people ingest marijuana?
Generally speaking, marijuana is ingested in two ways: smoking,Or just pick it up(No, it’s taken with food). When people smoke marijuana, the chemicals in it enter the bloodstream from the lungs and transport them to all parts of the body, including the brain. But if it is eaten, because it is absorbed through the digestive system, the effects of marijuana usually appear 30 minutes to 1 hour late.
For marijuana users, its most fascinating place is probably the euphoric relaxation after use. In addition, some people will also experience the erratic scene of sensory amplification, but some people believe that cannabis makes them feel anxious, fearful, distrustful and panic. Although there are currently few cases of death due to pure marijuana smoking, if overdose is used, cannabis intoxication (cannabis intoxication) is caused.[3]。
People with marijuana poisoning may develop symptoms such as hunger and lethargy. In severe cases, they may reduce their cognition and sense of orientation toward people and objects, and may even suffer from acute psychosis.[4].Other typical and predictable symptoms include dry mouth, red eyes, impaired short-term memory, and the effects of perception and movement, etc.5。
Some people with marijuana poisoning will have too slow response time to the outside world due to the mental impact of marijuana, causing poor performance at work or school, and even interference in driving and driving, which will eventually lead to traffic accidents, casualties and other regrets.
Although there are blood, urine or saliva tests that can detect cannabis poisoning, if you want toMoment monitoring in daily life, I am afraid there are still some restrictions.
In the past, there have been studies using sensors in the inseparable “smart phones” of modern people to detect high-risk drinkers with an accuracy rate of 90%.[6]. In view of this, the Rutgers University Institute of Health and Health Policy and Aging Research team began to study, wondering whether with the help of machine learning models, mobile phones can play a role in detecting “marijuana poisoning” and real-time detection of those that may be caused by cannabis poisoning. Crisis.
The team first recruited 57 young people between the ages of 18 and 25 from Pittsburgh, PA, USA. Through self-reports, they learned that they used marijuana at least twice a week. After that, the team used methods such as “mobile phone return survey” and “mobile phone sensor data” to collect daily data on subjects’ use of cannabis for up to 30 days to understand their status after cannabis poisoning.
Among them, the return survey is conducted three times a day, including the start and end of the use of marijuana, the amount of time, and the self-rating of subjective feelings-according to the degree of “Hi”, the scoring standard is 1 to 10 points, of which 10 is divided into “knock level” Hi”. Among the 451 marijuana use incidents reported later, the average “high degree” was 3.77 points.
The mobile phone is paired with an app to collect data from 102 types of mobile phone sensors, such as GPS, accelerometer, number of calls made, and average travel distance. Some people may not be able to sit still after hearing this. Wait…what can positioning tools and accelerometers like GPS do? In this way, GPS can be used to detect the travel boundary of marijuana users when they fall into “narcissism”, and the accelerometer is used to monitor their gait and physical activity.
After comparing the subjects’ return survey and mobile phone data, it was found that when the users reported that they were “positive,” through GPS data analysis, it was found that their range of movement was not far. In addition, the accelerometer data at this time also showed that the subjectively reported marijuana poisoning person, although the diversity of activities decreased, but the degree of physical activity was more intense.
Finally, with the help of algorithms[7][註1], I hope to know whether the above methods can distinguish between no poisoning and poisoning (mild or moderate). Through the behavioral characteristics of various poisonings and the verification of machine learning technology, the smartphone can become a fake “marijuana use monitor”!
In order to explore the accuracy of this combination, the team tried to investigate at different time points (for example: a certain day of the week, or a few minutes of a certain day) to find out the relationship between cannabis use behavior and a specific time point To further confirm the specific indicators of cannabis poisoning.
The results showed that only the sensor in the mobile phone was used to detect whether this group of people was using marijuana, with an accuracy of 67%; but ifCombine the “personal call and hemp time point” with GPS and accelerometer data,butUp to 90% accuracy。
Faced with such results, the research team believes that it is quite feasible to use mobile phones and machine learning to predict the degree of cannabis poisoning. However, more information needs to be added in the future to complete this tool.
First of all, because the study’s judgment on cannabis poisoning is mainly based on “subjective judgments and self（ㄕㄡˇ）Based on the “report”, therefore, the identification of substance use and physiological response is not as objective as the inspection tools of law enforcement agencies. In addition, things like the history of marijuana users, the route and dosage of ingestion, and the user’s tolerance to marijuana, will affect the results of their reporting of physical conditions.
Not only that, like when people who use marijuana infrequently are poisoned, is there any significant difference in their behavior and physical response compared with those “old drivers”? The subjects in this study are mostly white. Will other races produce corresponding data at the same dose?It’s not about race here, but just about “drinking”, there are some differences in the reaction of each race. It is generally difficult for Asians to metabolize alcohol.[8]。
All of the above are possible with this toolThe key to being universal. Finally, suppose this marijuana detection gadget is pushed to the app market one day, would you want to download it? Also, can we directly collect information and monitor cannabis users because of their sensitive identities and risks?As a side applaud, waiting for useful products to come outLittle people, While looking forward to the neck, we must also think deeply about such issues.
annotation
- Note 1: The technology used in this research is the “Light Gradient Boosting Machine”, which is based on the “decision tree algorithms” by Microsoft and released the LightGBM algorithm in 2017 for sorting , Classification and other machine learning tasks.
references
- Sang Won Bae et al. (2021) Mobile phone sensor-based detection of subjective cannabis intoxication in young adults: A feasibility study in real-world settings. Drug and Alcohol Dependence.
- How does marijuana produce its effects? National Institute on Drug Abuse, 2020.
- What are marijuana’s effects? National Institute on Drug Abuse, 2020.
- Helen Okoye. Cannabis Intoxication DSM-5 292.89 (F12.12). Theravive.
- Marijuana intoxication. U.S. National Library of Medicine.
- Bae et al. (2018) Mobile phone sensors and supervised machine learning to identify alcohol use events in young adults: Implications for just-in-time adaptive interventions. Addictive Behaviors
- LightGBM. Wikipedia.
- Hui Li et al. (2009) Refined Geographic Distribution of the Oriental ALDH2*504Lys (nee 487Lys) Variant. Annals of Human Genetics.
