New AI tool could help cops predict synthetic drugs before they hit the market

Synthetic drugs such as “bath salts” and designer weed show no signs of slowing down. The more governments try to identify and ban these substances, the faster new ones are invented and brought to market, even as overdoses continue to claim more lives.

Law enforcement agencies want to get ahead of the game and anticipate these drugs before they take off. So researchers at the University of British Columbia have created an AI model that can predict what kind of new synthetic drugs are most likely to be made and hit the market, giving cops a warning that could reduce drug investigations from months to days. The model is detailed in a new study published in Intelligence of natural machines.

The synthetic drug industry is essentially a form of advanced chemistry. Clandestine chemists develop and play with new molecules that mimic the psychoactive effects of more conventional drugs like cannabis. But moving around these molecules means the drugs aren’t technically the same as illegal substances, which means users get a legal high and manufacturers can’t be sued. The possibilities of new synthetic drugs are fundamentally limited by the imagination.

But AI technologies have the amazing ability to ingest massive amounts of data and produce more actionable analytics. In this case, the researchers fed an AI model with a database of hundreds of known psychoactive substances. The model has learned to predict 8.9 million potential synthetic drugs that could be developed, a testament to its capabilities.

The researchers then tested the AI ​​against 196 synthetic drugs that appeared on the market after the model’s training began. So these were drugs he didn’t even know existed. It turns out that the AI ​​had already predicted the emergence of more than 90% of these drugs.

Additionally, the AI ​​also learned to predict which types of molecules were more likely than others to appear on the illicit drug market. From the dataset of 196 new synthetic drugs, the model was able to correctly predict which chemical structures would be found in the top 10 drugs with 72% accuracy. Improvements to the model with other chemical data elements have increased this accuracy to 86%.

According to the authors of the study, some authorities around the world have already expressed interest in adopting and using the model in their investigations.

“There’s a whole world of chemical ‘dark matter’ just beyond our fingertips right now,” study co-author and UBC medical student Michael Skinnider said in a statement. “I think there is a huge opportunity for the right AI tools to shine a light on this unknown chemical world.”

Alvin J. Chase