Our newest ally in the fight against opioid abuse is a computer that’s learning how to figure out when humans are selling each other prescription medication on the internet.
A team of medical researchers in San Diego created an unsupervised AI to look for signs of illegal opioid trafficking on Twitter. After being fed tweets containing specific drug references, the machine was able to produce accurate data on opioid abuse by real-world location, and provide insight into online drug sales.
It analyzed over 600,000 tweets and found 1,778 cases of drugs being marketed online. The machine isolated the data geographically and was found to be consistent with comparable research conducted by humans.
The researchers concluded, in their paper:
Our methodology can identify illegal online sale of prescription opioids from large volumes of tweets. Our results indicate that controlled substances are trafficked online via different strategies and vendors.
It’s likely the research won’t be of any direct use to law enforcement, as the AI uses publicly available data to power its insights. Most criminals aren’t identifying themselves on Twitter while conducting illegal activity; this data is more important to researchers than police.
The more pertinent ramifications of this process are in its use as a tool to determine opioid abuse trends more efficiently than current methods.
Social media is an effective format for illegal prescription drug sales because it represents a (relatively) safe environment to conduct business, and access to a modest amount of anonymity. Moreover, it’s basically impossible for anyone to read every message on every network.
Even a clever computer program, designed to search for keywords, will yield unmanageable results. And those won’t be curated; you’re still looking for a needle in a haystack, there’s just less hay. Classical software solutions also aren’t capable of developing new indicators or fine-tuning its own methodology, AI is.
The researchers’ AI represents an end to traditional survey methods as scientific evidence. Instead of spending years looking high and low for verifiable data to try and paint a picture of how the opioid epidemic is growing, a machine can glance at Twitter and make the same determination — theoretically — in seconds.
Relying on records provided by patient care facilities and law enforcement agencies doesn’t provide information on the buyers and sellers who haven’t sought medical attention or been arrested — but Twitter does.
Instead of spending billions on the status quo, or cutting funding and suggesting methods that haven’t worked in the past, like “Just Say No,” it’s time we enlisted robots to help us make smarter decisions in the “war on drugs.”