Despite being a relatively new field, artificial intelligence is finding home in several industries including healthcare. Psychiatrists and psychologists have been using AI to effectively identify extremism or mental health issues in patients and treated them accordingly. But the footprint of drugs is traced beyond an individual’s addiction. When somebody is overly influenced by drugs, they tend to turn aggressive and end up attacking others. Therefore, in some cases, artificial intelligence in drug abuse is used to identify offenders over social media.
The healthcare industry is one of the early adopters of artificial intelligence. Even today, many institutes are using AI in rehabilitation facilities to help patients recover from alcohol or drug addiction. The National Institute on Drug Abuse reports that there are over 22.7 million Americans who need treatment for drug and alcohol abuse. Artificial intelligence is already revolutionizing addiction treatment by providing sophisticated and futuristic services. But before moving into the details, identifying drug abusers is the core of the following treatments. Generally, behavioral recognition technology is used to identify patients who need care. As computers become more adept at recognizing and predicting patterns of human behavior and emotions, the application of these predictions is becoming widespread. Starting from individuals who might attempt to commit suicide to people who could abuse others due to drug overdose, can be identified using behavioral recognition. AI could aid medical professionals to reach out to people who are in crisis due to drug abuse. Besides, artificial intelligence in drug abuse is helping minimize the chance of a relapse.
Using Self-taught Deep Learning to Identify Drug Abuse
Drug abuse is evolving to be the worst public health problem in the United States. Therefore, researchers have been looking for disruptive ways to identify drug abuses or potential drug abuses. As a result, they have picked the famous social media platform Twitter to help monitor individuals who might cause chaos. They also moderate drug abuse incidents over the social media platform. The researchers use an individual’s tweet to see if he is under drug abuse or not. Previously, the technology was not sophisticated enough to identify specific tweets based on their drug level. Usually, tweets are already noisy and sparse and the availability of labeled data is limited. Therefore, it was hard for scientists to find drug abuse cases.
Fortunately, a group of researchers is using self-taught deep learning systems to detect and monitor drug abuse risk behaviors in Twitter by using a large amount of unlabelled data. The model’s automatically augment annotated data improves the classification performance and captures the evolution picture of drug abuse on social media. To be even more precise, it is also giving the geographical information of the drug abuse-related tweets.
The research has used over three million drug abuse-related tweets with geo-tags and perform quantitative analysis to gain insight on drug abuse risk behaviors. The deep learning model identifies drug abusers’ tweets based on the words they use. They even extract local posting times based on their geographic locations to moderate positive and negative tweets. The geo-location information tagged in tweets is very helpful for capturing the distribution of drug abuse risk behavior.