The pandemic made room for technological advancements in healthcare. COVID’s strain on frontline workers, and on existing systems that have been used to treat, assist and service patients, spurred the development of new tech to assist and accommodate complex job demands.

Tech like AI chatbots and ChatGPT can handle tasks like, among other things, responding to insurance claims, which allows physicians to spend their time on patient-centered tasks. Because of the potential of AI, it’s tempting to believe that at some point it will be able to complete tasks without the need of human effort or input.

But artificial intelligence will never eliminate the need for pharmacists, even with recent developments in AI’s capabilities, said Samm Anderegg, CEO of DocStation, a pharmacy information technology company.

Healthcare IT News sat down with Anderegg to learn exactly why he believes this is the case, and what the future holds for AI and the pharmacy.

Q. You advocate for how technology can enhance healthcare, yet you caution against healthcare professionals putting too much stock in artificial intelligence. What's the balance you're trying to strike here?

A. Artificial intelligence has amazing potential to change nearly every industry imaginable. Right now, we’re seeing the power of consumer tools like ChatGPT that are making our day-to-day lives easier by digesting massive amounts of data on the internet to answer complex questions and draft useful synopses, articles, emails and itineraries.

We’re also seeing AI playing a role in businesses, taking minutes for virtual meetings and suggesting targeted ads based on our interactions with social media posts and content.

The Silicon Valley ethos of “move fast and break things” is great for building marketing or e-commerce companies. But when you break things in the context of healthcare, the consequences could be detrimental, such as making a misdiagnosis, causing medication errors, or, the gravest of concerns, putting a patient’s life at risk.

Getting it wrong could lead to patient harm, adverse outcomes or higher costs. What’s exciting is when we do get it right. There are so many manual, laborious, administrative tasks in the healthcare process that bog down our ability to maximize time on the things that we as clinicians can do and only what we can do.

Much like it’s helping make our lives at home and at work easier, AI could reduce or, better yet, eliminate the menial, non-clinical tasks that are currently part of our day-to-day work so we can spend more time on the things we love and we’re trained to do – taking care of patients. Leveraging AI to augment patient care and improve the patient/provider experience is the holy grail.

However, we must be cautious, careful and responsible about how we leverage AI compared to other industries in order to mitigate risk and potential for harm.

Q. You warn healthcare executives not to think that AI can replace pharmacists. Why can't AI accomplish such a feat?

A. There are several traditional pharmacy functions that artificial intelligence has the potential to replace. What people tend to miss is that the role of the pharmacist on the healthcare team is rapidly evolving.

At the turn of the century, most vaccines were delivered at primary care clinics. More than 20 years later, during the pandemic, more than three-quarters of people now receive immunizations at pharmacies.

Healthcare is rapidly shifting to convenient and accessible retail healthcare locations, and pharmacies are at the forefront of this transformation. In response, the pharmacy practice model is evolving to be more service-oriented.

Pharmacists are now taking more of a mid-level provider role on the healthcare team and are in an ideal position to solve access to care issues, especially in rural and underserved areas. With that said, patients value the human touch provided by pharmacists, just as they do their PCPs.

While AI may be able to provide information, it cannot provide the same level of understanding and problem-solving that a human pharmacist can. Using this frame of reference, we can think about AI impacting the role of pharmacists the same way we think about it impacting the role of primary care providers.

While AI has the potential to augment pieces of the care delivery process, patients still need a healthcare professional to communicate, educate and guide them on their healthcare journey. Pharmacists are often the most accessible, frequently visited provider on the healthcare team.

Patients and the healthcare industry as a whole will continue to rely on pharmacists to play the role as a frontline care provider and trusted patient advocate.

Q. So, what can AI do to help pharmacists do their jobs better and more efficiently?

A. Like other healthcare providers, artificial intelligence has the potential to make pharmacists' jobs easier and better by automating administrative tasks and augmenting the patient care process.

The first potential application is making administrative tasks and functions often delivered by support staff both faster and easier. During the pandemic, many pharmacies adopted software for scheduling appointments for immunizations and other services.

There are now several AI-assisted chat services that can assist patients to help with nonclinical functions like scheduling appointments, initiating medication refills and reaching out to follow up with patients.

As pharmacies evolve from dispensing medications to providing services, there are many applications around the medical billing and revenue cycle process that could be improved by AI. When billing is performed manually by support staff, errors occur, resulting in denials or delays in reimbursement.

Automation can help eliminate errors by flagging incomplete or inaccurate information before a claim is submitted. This can reduce the number of denied claims, make the billing process more efficient and maximize potential revenue.

Leveraging these tools to perform administrative tasks is generally low risk, but helps make extremely inefficient and burdensome processes easier, without the risk of causing errors or patient harm.

Pharmacies are also in the business of collaborating with payers in value-based arrangements to improve clinical outcomes.

While several software vendors and tools leverage basic logic (for example, if/then decision trees) to identify at-risk patients, AI has the potential to predict patients soon-to-be at risk. For example, if a patient has been filling medications regularly and starts picking their medications up later than expected in consecutive months, AI could identify this trend towards non-adherence, summarize the information, and present it to the pharmacist for evaluation and patient outreach if warranted.

Applications such as this could prevent an adverse event, improve patient outcomes and save the healthcare system money.

Just as AI reads large amounts of information and builds a vacation itinerary, it could also be used to summarize medical information such as diagnoses, allergies, lab values, treatment history, progress notes, etc., to automate encounter documentation, identify potential diagnoses or recommend possible treatments.

A relatively benign example would be recommending the most affordable medication option considering the medication formulary used for a particular patient's pharmacy/medical coverage.

While leveraging AI for clinical use cases would be a significant leap forward, there will likely be information that is not connected to or considered by the AI that could lead to risk.

This is why there remains a need for pharmacists and other providers to evaluate recommendations for safety, efficacy, appropriateness, from a holistic approach, before implementing a recommendation.

These examples merely scratch the surface on how AI could make pharmacists' jobs and day-to-day responsibilities easier, more efficient and more fulfilling in the near term.

Q. What are some of the legal and ethical considerations that pharmacists contend with that AI would have trouble managing?

A. The first legal and ethical consideration that pharmacists and other providers should be aware of around the potential bias identified with use of artificial intelligence in the healthcare setting. AI algorithms are trained using historical data, which can be inherently biased due to lack of access to healthcare in underserved populations.

Without adequate access to healthcare, underserved populations may be underdiagnosed or not receive the same treatment, compared to other populations with the same medical risk profile. It’s critical for pharmacists and other providers to understand how AI algorithms are developed and trained, and how quality assurance is performed, to understand potential bias.

It’s also extremely important for providers to understand the risk for error in clinical applications of AI. While AI has the potential to make clinical recommendations, algorithms may not have access to critical information needed to make “informed” decisions about treatment regimens.

One serious adverse outcome could be providers relying heavily or even blindly on AI to make treatment decisions. Our ethical responsibility as providers is to “do no harm.” It’s important for pharmacists to understand the potential for harm when relying on AI to augment clinical decision making.

One more consideration that may not be as apparent is patient consent. Using AI, especially in the healthcare setting, is still highly experimental. Thinking about it in context makes it easy to draw a comparison to the clinical trial space.

Just as patients understand the potential risks and benefits of participating in a clinical trial involving a novel treatment, patients and their providers must understand the risk and benefits of using AI as part of their healthcare experience. The technology industry has been known to be a generally unregulated space based on the rate of innovation, compared to the policymaking process. (Think cryptocurrency regulation in the finance industry.)

In order to “do no harm,” it’s part of our responsibility to adopt the same informed consent practices with AI as we use in other areas of the healthcare industry.

Follow Bill's HIT coverage on LinkedIn: Bill Siwicki
Email the writer: bsiwicki@himss.org
Healthcare IT News is a HIMSS Media publication.