Researchers from the University of Michigan in Ann Arbor studied whether AI could supplement pharmacists’ expertise in spotting pharmacy errors by performing automated checks to verify pills’ shape, color, and unique, 10-digit National Drug Code.
The study, published July 27, analyzed 65,274 images of 345 different medications in bottles from an unnamed mail-order pharmacy. The research team designed a model to identify each medication’s National Drug Code, which accurately predicted which drug it was presented with about 98.5 percent of the time.
