Unlocking the Potential of Artificial Intelligence in Health Care: A New Study

It is impossible to avoid the topic of artificial intelligence (AI) right now, and the integration of AI in health care is a groundbreaking innovation that has the potential to revolutionize the industry.

Recently, a proof-of-concept study by Ruth Carrico, PhD, DNP, MSN, FNP, a professor at the University of Louisville, and an independent infection preventionist (IP) consultant, delved into the transformative role of AI in identifying health care-associated infections (HAIs).

In an exclusive interview, Carrico sheds light on the challenges, triumphs, and key insights gleaned from this research she conducted with Timothy L. Wiemken, PhD, MPH, CIC, senior director, Digital Platforms team lead for Pfizer, and director of data science and epidemiology, at Saint Louis University Institute for Vaccine Science and Policy.

ICT's Monthly Topic: Patient Safety

The study aimed to explore how AI could revolutionize infection surveillance—a cornerstone of IP activities. Carrico emphasizes that surveillance serves as the foundation for understanding risk factors, crucial for devising effective infection control and prevention strategies. However, the labor-intensive nature of surveillance, in terms of both time and resources, spurred the investigation into AI's potential as an invaluable ally in this critical domain.

“Dr. Wiemken and I were interested in how you use artificial intelligence, particularly in identifying HAIs,” Carrico told ICT. “Now AI is at this point, we're not at the point where AI can pull together yet all of the different pieces of information that are used for diagnosis. But we are trying to learn in the proof of concept with this particular study was to see where are we with AI and looking at surveillance, and assisting the IP. We know surveillance is incredibly time-consuming, not only the amount of time but the resources that are involved in performing surveillance and surveillance is the mainstay of our activities because we use in our knowledge to identify risk factors, and those then are the things that we want to try to control. So, the study is focused on whether we can take the definitions that have been provided by the CDC National Healthcare Safety Network (NHSN) and use AI to respond or react to situations or scenarios that we pose back to the AI agent."

Carrico told ICT the key findings from the study regarding the performance of AI-powered tools in identifying CLABSI and CAUTI infections. “We were pleased because we found out that if we do a good job curating the document, that is, we give the AI agent the truth. So, like what constitutes an infection, and then we structure our questions in a format that will be able to be answered by the information provided, then it worked very, very well."

Carrico continued. “When I mentioned the importance of having the questions set up so the agent can answer them when we talk about retract and infection, we know that there are several different levels or different types of urinary tract infection. One would be an ABUTI [asymptomatic bacteriuria UTI], and one might not be an ABUTI. We are looking at the AI agent being able to quickly describe or go through the truth of the information when we asked it a question to categorize which type of CAUTI was that? We learned that as we formulate the questions that we used to template."

(Quotes have been edited for clarity.)