Commentary
Video
Author(s):
To safely and successfully use artificial intelligence (AI) tools in the pharmacy, credentialing bodies and professional associations will need to validate these tools for the individual pharmacist, said Harry Travis of The Travis Group.
Credentialing bodies and professional associations will play a crucial role in validating new artificial intelligence (AI) tools for pharmacists, explained Harry Travis, president of The Travis Group. Additionally, the fragmented nature of pharmacy data will require collaborative efforts to establish standards and improve data access for effective AI implementation.
Transcript edited for clarity; captions were auto-generated.
Transcript
How can pharmacists ensure AI tools remain accurate and free from bias in real-world use?
I think the pharmacists have to look to their professional organizations and credentialing organizations, and they're going to have to do the hard job of validating these tools. This is probably an overly simplified metaphor [or] comparison, but if you go way back in time, 30 years [or] 40 years ago, automatic pill counters came online. Did we trust that these automated pill counters are going to count out 90 pills? They all had to be validated, and individual pharmacists didn't do that. The accrediting bodies did it. [It’s going to be the] same thing with AI tools. The accreditation bodies [and] professional associations need to step up and fill that gap and communicate [to] the pharmacist that “We got your back on this.”
Data and access to data have been a continual struggle for health care. How do we ensure AI has access to the data it needs to be used in accurate, effective, and safe ways?
Data exists in so many different places behind so many different access codes and firewalls and things like that. Every company that exists in the data chain of pharmacy—and I'm in this business kind of on a regular basis—we really just have to go through the hard work of each tech vendor [and] each data provider trying to figure out, how are the APIs [application programming interfaces] getting set up? And do we have to come up with new industry consortia to help on data interoperability and data standards? Or [do] the ones that currently exist need to be adapted for AI so that the individual user has to feel confident that they can get the data. Because you’ve got the smartest agent in the world, but if it can't get the data, it's dead on arrival. And [it requires] teamwork, lots of teamwork.