How To Use Artificial Intelligence To Be A Better Clinical Pharmacist

In this article a clinical pharmacist with experience and knowledge about using artificial intelligence describes how it can be used to be a better clinical pharmacist. 

Authored By: Steven Smoke, Pharm.D., BCIDP, BCPT

Article Posted 18 March 2024

In healthcare’s evolving landscape, artificial intelligence (AI) promises to redefine pharmacy. AI chatbots like ChatGPT and Gemini offer not-so-subtle hints at AI’s transformative potential. But while speculation about potential future applications to pharmacy is easy, identifying practical applications today is more challenging.

Andrew Ng, a leading AI expert, suggests viewing AI as a tool for enhancing tasks, rather than replacing jobs. Pharmacists aim to ensure medication safety and efficacy—a goal that involves diverse tasks across different settings, often requiring skills beyond traditional pharmacy knowledge. Only by considering all these tasks can we identify opportunities to leverage AI’s potential. 

To help create awareness about application of AI to pharmacy practice, in this article I will review five areas where clinical pharmacists can use artificial intelligence to advance their practice.

1. Completing Writing Tasks

Pharmacists excel in medication management but receive less training in writing, a skill crucial for common tasks in clinical pharmacy – emails, policies, manuscripts, evaluations and more. ChatGPT, Microsoft Copilot and others are powerful tools to support (not independently complete) these tasks. In fact, a randomized controlled trial of professionals conducting work-related writing tasks found those randomized to work with an AI chatbot spent 40% less time on the task and output quality improved by 18%.

Yet, their use introduces “friction” due to the inconvenience of integrating AI tools into daily workflows. My approach is to use them when I am spending more time thinking about how to say something, than what to say. In the case of this article, the ideas are my own, but credit goes to ChatGPT Plus for editing for clarity and writing quality.

Regarding friction, keep in mind this is a temporary phase. Copilot is available, for a price, within Microsoft 365, their suite of Word, Powerpoint, Excel, etc. The same will apply for other tools and this adds tremendous value. When they are available on your hospital’s computers though, that is a different question altogether.  

2. Ctrl+F, but better

Believe it or not, the reliable Ctrl+F command, cherished for bringing simplicity to document-searching, has been eclipsed by the advancements of AI. Ctrl+F can struggle with vague or common terms, making it challenging to find specific information in dense documents. AI tools offer a solution, allowing users to easily extract precise information from lengthy texts with a simple conversation.

Not interested in reading that 77-page document from the NHSN on AUR? No problem. Simply ask Copilot for the details you need and receive concise answers with direct links to relevant document sections. This approach saves time and increases efficiency in information retrieval.

Importantly, friction is not a problem here. While I used to roll my eyes when Microsoft Edge wanted to be my default app for reading PDFs, I welcome the opportunity now. Open the PDF, click the “Copilot” button on the toolbar and you are off! This is referred to as retrieval augmented generation, or “RAG” for short. Notably, chatbots used in this capacity are less susceptible to problems of hallucination, given their concrete foundation, a singular defined document to serve as its source of truth.

This innovative AI search tool marks a significant advancement, enabling pharmacists to quickly access necessary information within extensive documents, enhancing productivity and accuracy in their work.

3. Presentations, data visualization and data analysis

The laborious task of converting a set of ideas into a visually appealing presentation is another task ripe for AI support. Powerpoint Designer, within Microsoft’s Powerpoint application is a frictionless tool that takes the work out of slide formatting. Enter a few bullet points and drop an image on the slide and let Powerpoint Designer do the rest. It will offer several elegant arrangements, eliminating the tedious task of resizing, rearranging and formatting previously needed to create visually appealing presentations. Other AI tools can also help generate a slide deck from a simple prompt. See Figure 1 below, a slide from Plus AI based on a simple prompt (Make a presentation on the impact of AI on pharmacy). Maybe not a finished product, but its a great head start. A quick Google search reveals many such tools.

FIGURE 1: Example of slide design using AI tools

In the domain of data analysis, ChatGPT Plus showcases a significant evolution. It translates your English language request into Python code, performing tasks ranging from creating data visualizations to conducting statistical analyses. As a constrained translation task moreso than an open-ended generation task, this approach may be less susceptible to hallucination. While users should validate these outputs for accuracy, the potential for broadening access to sophisticated data analysis tools is immense.

Tableau, a leader in data analytics, incorporates AI to enhance its capabilities. Its Explain Data feature uses AI to explore datasets in depth, offering explanations for anomalies and expanding users’ understanding of their data. See a demonstration here. While it is not a tool used widely among pharmacists, this example underscores the cutting edge of AI applications in analytics and visualization, indicating a shift towards democratizing these complex tasks. By making advanced capabilities available to more users, including those without specialized software knowledge, AI is enabling richer, more informed decision-making.

4. Drug references & literature review

AI is transforming the pharmacists’ approach to essential tasks like using drug references and literature reviews. Micromedex’s Assistant highlights this shift, enabling natural language queries with a drug reference that starts like a conversation with a colleague. While this is not yet a dramatic change in drug reference interaction, it is an incremental change from traditional keyword searches and hints at a future where interacting with digital resources feels more like dialogue than data retrieval.

Additionally, AI tools such as ChatGPT Plus, Scite AI, Elicit, Perplexity and others aim to overhaul the process of navigating scholarly databases like PubMed. These innovations propose a more streamlined approach to literature reviews. I asked each of them about the role of combination therapy for invasive MRSA infections. My assessment is that they are at a resident level; key considerations were identified but some of the latest and highest quality evidence is missing.

FIGURE 2: Example of using AI to answer a clinical question. Response from Consensus AI, a GPT available with ChatGPT Plus on combination therapy for invasive MRSA infection.

While pharmacists may have concerns about their reliability and the impact on our role, we have a unique opportunity to embrace these AI technologies. As trusted advisors on drug information, exploring these tools allows pharmacists to guide healthcare professionals and students towards effective resources. By vetting the efficacy and reliability of AI tools, pharmacists can ensure the continuation of high-quality care and maintain our status as indispensable resources in healthcare.

Adopting AI for drug references navigation and literature reviews can enhance pharmacists’ capabilities, enabling them to navigate an increasingly digital information landscape efficiently. As these technologies prove their worth, they could reinforce pharmacists’ roles, blending traditional expertise with cutting-edge tools to better serve patients and the healthcare community.

5. Chart Review Gone Wild

Chart assessment, a crucial yet time-intensive task for clinical pharmacists, is invariably inefficient in its impact on patient care. What if AI could handle the initial chart surveillance, allowing pharmacists to focus on the tasks that impact patient care, the actual interventions? This shift could dramatically enhance efficiency and enable pharmacists to extend their care to more patients.

Technologies such as Sentri7’s Sepsis Monitor, which uses natural language processing (NLP) to identify patients with sepsis by reading charts, showcase the potential of AI in clinical tasks. Applying similar AI tools for antimicrobial stewardship or other clinical pharmacy-related areas could significantly expand the reach and effectiveness of pharmacists’ work.

Adopting AI for surveillance tasks or retrospective medication use evaluations promises to dramatically streamline clinical pharmacy practice. By concentrating pharmacists’ expertise on critical interventions rather than preliminary assessments, this approach could greatly improve patient outcomes and optimize pharmacists’ impact on healthcare.

Concluding thoughts

While AI tools offer transformative potential for pharmacy practice, it’s crucial to approach their integration with awareness of limitations, concerns, and the need for critical evaluation. For more information on these issues, see here. 

Understanding and engaging with these AI applications, even through limited means like free trials or demos, is vital for pharmacists. Such engagement not only aids in grasping the evolving landscape but also prepares pharmacists for future roles that leverage technology for enhanced patient care.

Despite the enthusiasm for AI’s potential, practical barriers like friction in tool integration can temper immediate adoption. As an AI optimist myself, I recognize that while I do not use these tools daily, their growing accessibility is set to redefine our professional practices significantly.

Rapid advancements in AI capabilities further underscore the need for pharmacists to remain informed and adaptable. Resources such as the NEJM AI Journal, NEJM AI Grand Rounds podcast, and Hard Fork podcast offer valuable insights into these developments. The experience of using AI tools is constantly evolving, emphasizing the importance of staying engaged with these technologies.

By embracing AI, pharmacists can ensure they operate at the forefront of efficiency and quality, fundamental components of our mission to improve patient care. As we navigate these changes, staying informed and adaptable will be key to harnessing AI’s full potential in enhancing the pharmacy profession.

REFERENCES/ READINGS

1. Noy S, Zhang W. Experimental evidence on the productivity effects of generative artificial intelligence. Science. 2023;381(6654):187-192. doi:10.1126/science.adh2586

2. Soudani H, Kanoulas E, and Hasibi F. Fine Tuning vs. Retrieval Augmented Generation for Less Popular Knowledge. arXiv. 2024 Mar 3;arXiv:2403.01432.

3. Riley D and Chiang D. A Continuum of Generation Tasks for Investigating Length Bias and Degenerate Repetition. arXiv. 2022 Oct 19;arXiv:2210.12929.

DISCLOSURE: The views and opinions in this article are that of the author and do not necessarily reflect the opinion or policy of any former, current, or potential future employer.

ABOUT THE AUTHOR

Steven Smoke received his Doctor of Pharmacy from Rutgers University and completed a PGY-1 pharmacy residency at Penn Medicine Princeton Medical Center. He worked as an antimicrobial stewardship pharmacist for 10 years and today is an informatics pharmacist at Newark Beth Israel Medical Center in Newark, NJ.

Dr. Smoke has authoring several peer-reviewed publications on antimicrobial stewardship, antibiotic resistance and pharmacokinetics. His interests include antimicrobial pharmacotherapy, informatics and artificial intelligence.

RECOMMENDED TO YOU