APhA 2025: Enhancing Medication Management and Patient Safety With AI

Pharmacy Times® interviewed Ravi Patel, PharmD, MBA, MS, lead innovation advisor at the University of Pittsburgh School of Pharmacy, on his presentation at the American Pharmacists Association (APhA) Annual Meeting & Exposition.

Patel discusses the applications of artificial intelligence (AI) in pharmacy. He highlights 2 types of AI technologies: natural language processing for analyzing patient information and computer vision for tasks like pill counting. Patel explains how AI improves medication management by dynamically analyzing data to identify potential drug interactions and assess the applicability of guidelines to individual patients. He also emphasizes AI's role in enhancing patient safety by detecting long-term risks and patterns in medication use that humans might miss.

Pharmacy Times

What are 2 types of AI technologies used in pharmacy today?

Ravi Patel

Artificial Intelligence in pharmacy today are examples of new technology and old and existing technology. For example, in the past, we've used natural language processing to analyze information like a SOAP note or structured data like finding drug interactions. With new technology, concepts like computer vision have gotten so accessible that even our cell phone could be used to instantaneously provide a count of pills that are on a counting tray, so that if I remove a pill, I immediately know if I go from 35 to 33 to 30 pills all in the same instant. So that's made my workflow much easier.

Pharmacy Times

What is an example of how AI is improving medication management?

Patel

AI, in the frame of medication management, has a lot of potential opportunities. Some ways that it's already being used, often include concepts like drug drug interaction, while there's some existing rules-based approaches for identifying drug interactions, some exciting ways that dynamic data collection and analysis have been afforded by artificial Intelligence. For example, being able to look at a specific set of guidelines and comparing that to a list of medications can help identify if the information or conclusions from that paper might apply or might not apply to a given list of medications for a patient.

Pharmacy Times

How can AI enhance patient safety in a pharmacy?

Patel

When pharmacists think of their core skill set, there's a framework of using medication safety and medication efficacy. With medication safety, being able to identify drug drug interactions have been part of static data, a list, compared against another list. Some of the unique ways that artificial intelligence is building into drug safety includes identification of risks that might develop over time. If we know, for example, that a specific medication might have been discontinued or stopped, and we see a trend of those medications that are stopped automatically, artificial intelligence or these advanced computing technologies might be able to identify a pattern that any one human being able to process or identify that pattern by their own brain might not have been able to do without the help of some technology.

Pharmacy Times

Can you describe AI's impact on personalized treatment plans.

Patel

With artificial intelligence, personalized treatment plans might be possible for more local data. An example of that might be being able to use the population of patients within a small clinic or within a single pharmacy to identify local impact for the antibiogram or the efficacy of an antibiotic treatment for the strains of something like community acquired pneumonia within a specific geographic region. So, when we think about the impact of artificial intelligence and personalized treatment plans, we have the ability to reconcile dynamically new and updating data with the changes in practice that might be able to make an impact from that data.

Pharmacy Times

What are potential challenges of implementing AI in a pharmacy setting?

Patel

The challenges from artificial intelligence in pharmacy settings can be multifactorial. When we think about the risk of generated data, hallucinations often come to mind. Information that's generated that might not have any basis in reality, can be a major risk when we're thinking about identifying our evidence or what conclusions or treatment pathways, we might be able to take. By being very thoughtful of how we understand our data or validating any kind of output from generated artificial intelligence answers, we can apply that medication expertise that pharmacists have into this new field of Artificial Intelligence. Other ways that pharmacy might need to adapt to these new technologies include trust in the technology by understanding low risk and low outcome assessments that might be able to validate the role and impact of artificial intelligence, we can start to build not only our trust, but also our understanding. And what are these technologies able to do, and where might they not fit in. So, as an example, we might think of drafting a care plan or a letter to a physician that still allows the human and pharmacist expert to create the final version that's eventually documented or submitted.

Pharmacy Times

Is there anything you would like to add?

Patel

Artificial intelligence has been around for longer than we might think about. Being able to use a decision tree asking yes or no or ruling out different care. Pathways are ways that we've used to create algorithms with our human minds. By using a large amount of data, these artificial intelligence pathways allow us to find new ways of thinking about care that might not have been possible before. So, from 100 years ago, where the concepts of AI were started, to 50 years ago where we started seeing these rules-based decisions, it's really exciting to see, just in months, how quickly the technology evolves. The integration of artificial intelligence often reflects how we might think of a lot of other technologies just in a different time frame. As an example, being able to use Google for drug information evolved over time so that we can refine our own medication information seeking behaviors to online resources and libraries named specifically for that fact, the same way we use artificial intelligence to find potential answers will refine over time. Already, we see examples of medication information specific bots and platforms for artificial intelligence.