HOW IS AI HERE, THERE, AND EVERYWHERE, BUT NOT IN your pharmacy? Even though AI (artificial intelligence) surrounds us in nearly every aspect of our lives, including the smartphone in your hand or pocket right now, it’s been slow to be used by pharmacies.
Whenever I find myself in a pharmacy I haven’t visited before, I ask the pharmacist if AI is doing anything to help them in their job. I always get the same answer, whether it’s a large chain or a small independent pharmacy, “No!”
When I tell them my role with DrFirst is about bringing AI to help pharmacies improve efficiency and safety, the reactions range from curious to skeptical. It’s hard to blame healthcare providers for being skeptical about AI. After all, when lives are at stake, it can be hard to imagine trusting AI with healthcare data when we are so used to AI’s shortcomings in everyday life. How many times have you laughed at the ridiculous auto-correct fails on your texts, for example?
Yet, AI has made such tremendous progress that most people don’t even consider how often we use it. Twenty years ago, my finger pressed each button on a phone attached to my wall to place a call. Now, I tell my phone who to call.
In my previous article in the September/October issue, I shared how clinical-grade AI differs from the AI on your smartphone, and that it’s been successfully used by hospitals and clinics across the United States and Canada since 2015. Today, through integrations with industry-leading electronic health record (EHR) systems, clinicians leverage AI for over 24 million medication transactions a day.
Clinical-grade AI has two powerful benefits for busy pharmacies. First, it’s a “force multiplier,” quickly and automatically codifying data as it comes into your pharmacy management system (PMS), which frees up your staff for more productive and rewarding work. For readers who don’t speak geek, codifying is taking what we see as typed text and interpreting, arranging, and assigning it into fields in a computer system. Doing so enables the system to perform calculations like verifying a dosing for safety.
Second, clinical-grade AI is a “safety multiplier,” working as a co-pilot to staff and reducing the opportunity for keyboard errors. In complex cases, the AI flags pharmacy staff whenever their judgment is needed.
In the United States, where electronic prescribing has been the norm for many years, pharmacy technicians still manually enter data for nearly every prescription, reading every order to understand the clinical intent and then typing each piece of information into the various boxes in the PMS workflow. That’s because the computer systems don’t all speak the same language. A doctor’s EHR sends out a prescription using its native language, and the PMS doesn’t understand it. As a result, much of it arrives as a block of free text because the data doesn’t fit the discrete fields in the PMS.
Clinical-Grade AI for Clean, Consumable Data
Clinical-grade AI can speak both languages, resulting in near-seamless communication. The AI:
- Matches a prescription’s drug description and NDC (National Drug Code) with correlating drug identifiers that are useful to the PMS beyond its local drug database system, resulting in the drug matching directly to the pharmacy’s preferred drug over 96% of the time.
- Translates up to 93% of free-text sigs into dose, dose unit, route, and frequency values, coded into the specific receiving pharmacy’s unique short codes.
- Links 97% of allergy information from the prescriber’s EHR into the PMS, allowing for early triage of potential safety issues before dispensing.
The AI saves countless clicks and keystrokes by providing clean, consumable data to the PMS. The pharmacy staff reviews the output against the original order and makes the final call based on their clinical judgment. A national mail-order pharmacy saw a 300% increase in prescriptions processed and an 80% decrease in transcription errors. A small, independent pharmacy saw a 32% decrease in time spent on data entry.
Examples of Transformation through Clinical-Grade AI
In Canada, where e-prescribing is not yet mainstream, AI is making a transformative difference for pharmacies. Currently, nearly all prescriptions are faxed to the pharmacy, showing up in a fax queue as unlabeled documents in differing formats. A busy pharmacy typically receives several hundred faxes a day, and patients commonly arrive before the prescription is ready. The staff must then sift through the faxes one by one, looking for a needle in a haystack. Next, they must manually enter all the faxed information into appropriate fields in the PMS (patient name, date of birth, street address, city, province, provider name, medication name, strength, quantity, etc.). It takes time, and each step is an opportunity to introduce an error.
Six pharmacies in British Columbia are now using clinical-grade AI in a pilot program to auto-fill the data from the fax images into the distinct fields within the PMS, with the pharmacy staff verifying the data instead of typing it. The initial results are extremely promising, and a regional chain serving several provinces is beginning its pilot now.
If “expectations are the root of all heartache,” what might that mean for AI in pharmacy? For skeptics, like so many of us in healthcare, the key is knowing how clinical-grade AI has already proved itself. It safely streamlines workflows, prevents errors, and saves staff from mind-numbing, tedious data entry. What started in hospitals nearly nine years ago is now expanding into pharmacies. So, before too long, AI may be “here, there, and in your pharmacy.” CT
David Sellars, C.Eng., serves as principal, product innovations at DrFirst. He has over 20 years of healthcare experience with a deep emphasis in big data, AI, healthcare interoperability, programming, and systems optimization.