A pilot project using AI to strengthen patient safety offers insights into how organizations can integrate the technology into pharmacy practice.
At the upcoming ASHP Pharmacy Futures 2026 meeting, presenters of Reimagining the Medication Safety Role in the AI Era will share lessons from that project at University of California San Francisco (UCSF) Health.
Sylvia Stoffella, medication safety specialist at UCSF Medical Center, and co-presenter Benjamin R. Michaels, a UCSF Health data science pharmacist, gave the ASHP News Center a glimpse of what’s in store.
Augmentation, not replacement
The most successful implementations treat AI as a tool that supports pharmacist judgment rather than replacing it, said Stoffella. “AI is an enhancement,” she said. “The human element is always there.”
That dynamic carries a range of implications, the presenters said.
- Medication safety professionals must shift their focus from manual, reactive workflows toward proactive risk detection and intervention.
- AI supplementation is about more than getting the right data to the right users at the right time. It’s about outcomes.
- “We created an innovative AI driven process around it, but the end goal must be to do something with it and drive outcomes,” said Michaels.
Seeking clarity in care
The UCSF Defogger Program — its name a nod to the Bay Area’s famous weather — was developed to improve the cumbersome process of reading, categorizing, and evaluating medication safety event reports. The UCSF medication safety team reviews about 3,000 reports per year, a number that continues to increase as the health system expands.
The presenters plan to give an overview of how the project was designed and implemented.
- Defogger enhances UCSF Health’s existing reports with three new pieces of information for each medication safety event: contributing factors, corrective actions, and countermeasures, said Michaels.
- The program is especially useful when a medication safety professional is less familiar with an area of practice. “Defogger really helps provide guidance on what happened in this event and what some potential solutions could be, which we can then validate with clinicians,” Stoffella said.
- The Defogger pilot is still in the testing phase. Ideally, future iterations of this AI tool will help pinpoint medication safety trends at the department, location, and system levels.
- Stoffella also hopes to reinforce UCSF Health’s culture of safety. “Again, the ultimate goal is system change,” she said.
Looking beyond medication safety
The Pharmacy Futures presenters will also offer general guidance on innovating with AI.
- Stoffella said AI projects must involve collaborators from the start. Defogger was a partnership between the medication safety and pharmacy analytics teams, for example, with future goals to include heavy input from interdisciplinary care providers.
- She added that AI solutions must mesh with existing workflows. “If people don’t understand how to use it, then it’s not going to be useful,” cautioned Stoffella.
- AI-driven systems must be rigorously validated before they are put into practice, the presenters say, including testing the technology on historical patient data and piloting it in a controlled setting. Clinical pharmacists and physicians must continually monitor the tool’s output for potential issues.
- Before scaling any AI initiative, say Michaels and Stoffella, evaluate whether the technology has truly saved resources and meaningfully improved patient outcomes. If not, cut your losses and move on.
The speakers say they hope their case study will trigger new ideas among fellow meeting attendees.
“It will be exciting to see how this could be applied to another process that involves reading something on a daily basis, whether it’s clinical, operational, or financial,” Michaels said.
