Scott V. Anderson, director of member relations and the liaison for the section of pharmacy informatics and technology at the American Society of Health-System Pharmacists
Photo: Scott V. Anderson
Artificial intelligence's momentum is undeniable – but nowhere are the stakes higher than in healthcare, where patient safety and trust are on the line.
AI in healthcare requires proof of effectiveness, transparent models, reliable data, and, above all, clinical oversight, said Scott V. Anderson, director of member relations and the liaison for the section of pharmacy informatics and technology at the American Society of Health-System Pharmacists. He holds a pharmacy doctorate.
Hospitals and health systems must navigate these challenges, pitfalls they face, and steps to ensure AI improves care without sacrificing safety or trust, he added.
The ASHP, based in Bethesda, Maryland, is the largest association of pharmacy professionals in the United States, representing 60,000 pharmacists, student pharmacists and pharmacy technicians in all patient care settings, including hospitals, ambulatory clinics and health system community pharmacies.
We sat down with Anderson to glean his expertise on what the ASHP sees as bright spots of activity of artificial intelligence at health systems; how AI is affecting the day-to-day lives of pharmacists; what health systems should be looking for as proof of effectiveness; the transparent models and reliable data factors; and the requirement for clinical oversight.
Q. ASHP represents caregivers working at health systems. What is the organization seeing as bright spots of activity of artificial intelligence at these health systems?
A. Currently, pharmacists are expressing optimism toward the potential of AI tools and the ability to enhance both clinical practice and patient interaction. In particular, the ASHP Foundation 2025 Pharmacy Forecast Report contains some excellent insights into how generative AI is going to affect practice. The results show a sentiment among health system pharmacists that in the next five years the use of AI will have tremendous potential to expand patient access and increase the effectiveness of care.
In current practice, operational efficiencies are making the biggest impact. The time needed for tasks that are necessary but repetitive and take away from direct patient care, such as the processing of prior authorizations or received faxes, can be mostly automated using AI tools.
Clinical application of AI tools is where we're seeing the most excitement. The ability to combine efficiency and robust clinical knowledge is a game-changer for pharmacy practice. Tools such as AI scribes and clinical decision support are quickly gaining momentum.
Pharmacists are able to record notes and quickly digest multiple elements of a patient chart without having to search for the information, which allows them to spend less time in the electronic health record and more time with patients.
In the future, additional processes will benefit from the addition of AI tools. Ideally, using AI to automate repetitive manual tasks will help pharmacy professionals focus on patient-centric tasks with the most clinical impact. Also, the combination of AI with population health data will help us find the patients that most need interventions but are historically difficult to identify.
Q. How is AI affecting the day-to-day lives of pharmacists?
A. AI is providing pharmacists with more data, faster than ever, which is changing the paradigm of providing care. Using tools in the patient chart to compile data and create notes is helping with efficiency, but many pharmacists are using AI outside of the chart as well. AI tools that help locate and digest clinical data, perform drug information tasks, and provide outlines for communication and presentations are all having a positive impact.
Patients are increasingly using AI to assist in their own healthcare, and pharmacists are adapting their practice to help patients better understand the information they are reading and educate them how to know the information is trustworthy. Embracing the use of AI by our patients is important to make sure their health isn't compromised by incorrect information.
AI is also helping pharmacy professionals expand their role in the health system. Especially for AI tools that affect the medication use process, pharmacists and pharmacy technicians beyond the informatics teams are getting involved with evaluation, design, implementation and assessment. These tasks help with familiarity but also help identify when the tools are not performing as expected.
Q. You say AI in healthcare requires proof of effectiveness. What should health system executives and clinicians be looking for as proof?
A. Proper governance is essential in ensuring the safe and ethical use of AI in healthcare. Executives should establish interdisciplinary governance committees with oversight of AI tools, which will help establish baseline trust in the AI that is being implemented. These committees should create guidelines that properly assess if an AI tool is the appropriate solution as well as mechanisms for ongoing evaluation.
AI can't solve every problem or increase efficiency in every situation, so establishing realistic expectations and proper evaluation is crucial to success and ensuring the ability to measure effectiveness. A clearly defined end goal for using AI improves the ability to not only measure effectiveness but also communicate the results with executives who may not fully understand clinical processes.
Clinicians will also seek proof of effectiveness, not only for their practice but also to help maintain transparency with patients about the use of AI in their care. Having clinicians involved in the development and implementation builds trust in the system and deepens the understanding of how AI is working, which in turn helps with communication with patients.
Q. You also point to transparent models and reliable data. Please talk about these AI factors.
A. With data for healthcare AI coming from multiple sources, collaboration is key for reliable data. Health systems, vendors, payers, national databases and others all need to clean and process their data to ensure interoperability across the spectrum of healthcare. As the use of AI in healthcare grows, maintaining and improving data standards will help immensely in having reliable data.
Clinicians will want to know the source of the data they are using in their practice, as well as assurances that it is as complete and accurate as possible. With proper development and implementation, clinicians should be able to identify their data sources and make decisions on the trustworthiness of the data.
To maintain trust, assessments for bias and drift are key, especially for models that are continuously learning from new data. Clinicians need to be able to identify when they suspect bias and drift are occurring and have mechanisms in place address those issues.
Patients also have trust issues with AI, so we need to ensure clinicians understand and trust their data and can share that information with their patients.
Q. And you say, above all, AI in healthcare requires clinical oversight. Please describe, from your point of view, what best practices look like here.
A. Each clinician should be appropriately trained in the use of the AI tools they use. Understanding when it is working and when it isn't, having appropriate mechanisms for reporting issues, and maintaining a relationship with informaticists are all key elements to effective clinical oversight. Clinicians often experience the unexpected, especially during pilots or just after a go-live, so collaboration with the informatics teams is important.
Clinicians also need to have plans in place for maintaining their practice in the case of a downtime, down-scaling or roll-back of an AI tool. AI can help improve our standard of care, but providing the best possible clinical outcomes should still be possible when the AI tools aren't working appropriately or need to be removed altogether.
The most important part of clinical oversight is maintaining the clinician-patient relationship. Each clinician should be able to trust that AI is helping them provide optimal care, just as each patient should trust that AI is being used effectively and safely to improve their experience.
Adherence to ethical frameworks and the maintenance of AI source data and tools helps improve the experience and results of care for everyone in healthcare and their patients.
Follow Bill's health IT coverage on LinkedIn: Bill Siwicki
Email him: bsiwicki@himss.org
Healthcare IT News is a HIMSS Media publication.
WATCH NOW: Wisdom on nurses and IT from a health system's nurse CIO
