At McKesson ideaShare 2024, Andrea Sikora discussed artificial intelligence and how it has gradually been integrated into health care systems.
The genesis of artificial intelligence (AI) in the health care space has led to this type of technology being the core infrastructure for maximizing patient care. As health care technology evolves, it is predicted that AI will have a role in all new resources and tools introduced throughout the ongoing digital age.1 But before health care providers integrate it into their process, it’s important for them to understand how it works.
In a continuing education session, Andrea Sikora, PharmD, of the University of Georgia College of Pharmacy, introduced the basic functions of AI, then delved into the technology’s emerging role in health care and highlighted the key areas of utilization that must be addressed before integrating AI into the pharmacy. Sikora’s session was presented during McKesson ideaShare, held June 23 to 26 in New Orleans, Louisiana.
There are specific subsets of AI functionality that allow the technology to go above and beyond traditional mathematical algorithms, which have been the standard for technological programming before the introduction of AI.1 One of those subsets is machine learning. “Machine learning is essentially mathematics and statistics to do different types of modeling,” said Sikora.1 “[It] is the ability for the computer to learn without being explicitly programmed with a set of rules.”
To better understand machine learning, Sikora compared this subset of AI to traditional programming technologies through a simple traffic light example. Sikora explained that with traditional programming, technology is given directions to know that a green light means “go” and a red light means “stop.”
But with machine learning, the technology is set up in front of the same stop lights and it is programmed to teach itself. Eventually, the machine will observe traffic light activity and find that the majority of data report that cars go when the light is green, and stop when it is red. Put simply, machine learning is the ability to learn from a data set.
Andrea Sikora, PharmD, introduced the basic functions of AI then delved into the technology’s emerging role in health care. | image credit: Supatman / stock.adobe.com
Sikora dove further into machine learning and its 2 different types: supervised and unsupervised. Supervised machine learning is when technology is directed to look for specific data; unsupervised learning is a way for the machine to figure things out on its own, finding patterns within data sets rather than with a specified end goal in mind. Furthermore, supervised learning directs machines to search for binary outcomes and classification, while unsupervised learning is designed for AI to discover patterns.
“One of the biggest things is that pretty much every type of regression assumes a linear relationship of some kind, but not everything in healthcare, and particularly pharmacy has that linear relationship,” she said. “AI can handle that in a way that logistic regression cannot.”
READ MORE: NCPA Says AI Can Help Pharmacies Optimize Cash Pricing
Sikora then transitioned into AI in the pharmacy and the things pharmacists should consider before integrating AI tools into their daily operations, touching on vendor-introduced AI, the integration of AI in telehealth, and the legal concerns surrounding this new technology.
Vendor-introduced AI is artificial intelligence used by outside, third-party vendors to help, in this case, pharmacists streamline their processes. When working with vendors, Sikora listed a plethora of important questions to ask before committing to this outside technological assistance: “Can you tell me about the datasets that were used to train and validate this? Is this proprietary data? Is it available? Can I see the data dictionary? Can you tell me about the demographics? Do you relate to the demographics of patients that they use in their validation and training versus the patients that you take care of? Can you get results from other users?” said Sikora. “Can you talk to another user and say, ‘How would you use this and how did it go?’”
When it comes to the inception of telehealth in health care, Sikora highlighted the need for collaboration, not only between health care entities but between specific technologies that have grown within this space. The key collaboration in the telehealth space is integrating AI: If independent pharmacies are able to utilize AI to more easily streamline telehealth processes, everybody will benefit in the end, including patients, pharmacists, owners, and their profits.
As with any new technology, there are legal concerns surrounding AI, especially around the Health Insurance Portability and Accountability Act (HIPAA), state laws, and important liabilities that independent pharmacists must understand. As AI emerges and becomes commonplace in the health care industry, it creates further checks and balances that pharmacists must keep track of to get the most out of this technology.
Sikora concluded by sharing a few of the more important questions pharmacies should be asking about AI utilization. Among others, those questions are: “Are there real-world results for how AI is used in each specific situation?”, “Does the technology update as new data is unveiled within the space?” and “What are the input variables, or data dictionaries, used to come to conclusions?”
“All I can say,” Sikora said, “is stay tuned.”
Click here to check out more of our coverage from McKesson ideaShare 2024.
