AI tool designed to identify buyer behavior reimagined to discover medicine safety and repurposing opportunities

Monash University researchers are using the power of consumer AI to evaluate the safety and repurposing potential of medicines.

The ‘association discovery’ data mining tool, known as Magnum Opus and typically used to pinpoint buyer behavior patterns, has been redefined to uncover patterns in medicine safety and drug discovery.

The study is published in Clinical Pharmacology and Therapeutics and brings together medicine and technology expertise from the Center for Medicine Use and Safety (CMUS) within the Monash Institute of Pharmaceutical Sciences and Department of Data Science & AI at the Faculty of Information Technology (FIT).

The study used a 10% sample from the Australian Pharmaceutical Benefits Scheme (PBS) comprising over 300 million prescription records between 2014 to 2024 to identify associations between a wide range of medicines and three common chronic medical conditions: coronary heart disease, type 2 diabetes, and epilepsy.

The study analyzed all the medicines each cohort was prescribed before their condition, with the team able to spot and categorize several patterns. For example:

• Coronary heart disease: Expected links appeared with cholesterol-lowering and blood thinning medicines, possibly due to treating overlapping conditions.
• Type 2 diabetes: Some medicines (e.g. antipsychotics, diuretics, statins) were expectedly linked to higher diabetes risk, while others (for Parkinson’s or osteoporosis) were surprisingly linked to lower risk.
• Epilepsy: Many expected associations between antidepressants, antipsychotics and increased epilepsy risk were observed, while a common blood pressure-lowering medicine was unexpectedly linked to lower risk.

Dr. George Tan, a research fellow with CMUS and co-lead author, said the findings show how powerful prescription data can be in revealing the hidden connections between medicines and health outcomes.

“Medicine use patterns tell a story. By studying prescriptions, we can not only see how conditions are being treated, but also discover surprising links that may point to new risks, new protections, or even new uses for existing medicines,” Dr. Tan said.

Professor Geoff Webb from FIT, the senior author, said, “This study shows how big data can generate early clues about medicine safety and effectiveness in much the same way it helps predict consumer behavior.”

“AI is helping us move from reactive to proactive health care,” Professor Webb continued.

“By harnessing the same algorithms that predict what people might buy next, we can begin to anticipate which medicines may work best for which patients—and which might pose hidden risks. It’s about using AI not just to analyze data, but to generate new medical knowledge that can ultimately improve patient care.”

Dr. Lynn Miller from FIT, co-lead author, also mentioned, “By adapting association discovery algorithms to health care, we can uncover meaningful relationships across millions of prescriptions, not to replace clinical research, but to help guide where that research should look next.”

It’s important to emphasize that these findings are early-stage and hypothesis-generating. Further studies are needed to refine and validate the signals by testing their timing, biological plausibility, and consistency across different datasets.