STAT+: AI is good at catching drug theft at hospitals, but only when humans do their part

In late September 2024, a nurse at Adventist Health in Bakersfield, California, was observed behaving erratically by patients. The nurse was seen walking around the intensive care unit without shoes, talking to herself, and displaying aggressive behavior towards others. Concerned family members of patients noticed the nurse handling IV needles carelessly, but were hesitant to confront her.

One family member later revealed to federal investigators that they believed the nurse was under the influence of drugs. Another patient in the post-anesthesia recovery unit treated by the same nurse suffered severe pain under her care. The nurse, hired through a travel nursing agency several weeks prior to the incident, had been pilfering medications from a secured cabinet and administering them to herself, while falsely documenting that they had been given to patients.

The incident was not a random occurrence, as hospital managers had ignored alerts from a machine learning software designed to detect potential drug theft. Auditors discovered that the hospital had failed to address these warnings. Hospitals often hold large quantities of both addictive substances and necessary medical treatments.

Unfortunately, there are instances where employees steal these drugs and unknowingly pose a risk to patients by being under the influence while providing care. Although new technologies are aiding in identifying theft and connecting individuals with substance abuse resources, incidents like the one at Adventist Health demonstrate that artificial intelligence is not infallible.