Preventing Infections with Artificial Intelligence

May 28, 2024
AI remains a promising tool for clinical use, but manual review remains vital

Artificial intelligence (AI) is here to stay, and its ramifications are being felt across practically all industries around the world. AI in healthcare, however, is an even hotter topic given its proximity to decision-making that can literally be life or death for the patients involved.

Healthcare Purchasing News was able to talk to Isis Lamphier, MPH, CIC, manager of Infection Prevention at Moffitt Cancer Center in Tampa, Florida; and Jill Holdsworth, MS, CIC, FAPIC, NREMT, CRCST, CHL, manager of the Infection Prevention Department at Emory University Hospital Midtown in Atlanta, Georgia, about concerns, exciting developments, and next steps regarding AI in the infection prevention space.

Lamphier has heard “whispers of AI entering the infection prevention space,” including “upcoming presentations regarding AI at the annual APIC conference.” So far, she has seen that AI “can be used to aid infectious disease physicians” and “to provide models on the effect of infections and organisms in their patient populations.” Indeed, she says, “AI can be used to detect trends in HAI [healthcare-associated infection] surveillance and used for early identification of potential outbreaks and clusters of infections.”

Holdsworth also emphasizes the possibility of AI assisting with detecting trends in HAIs. She specifically feels that “some of the triggers we use to notify us of specific HAI criteria could easily be worked into an AI algorithm when scanning chart information.”

AI technology is promising in many areas, but it is certainly not in its final, most trustworthy state. Lamphier emphasizes that it is “important for IPs to always validate AI tools manually, especially at the beginning of this technology. IPs should verify the data and that patterns/trends truly exist before creating conclusions, [using] AI as a tool” and not the last word on a subject. She also says that it is ultimately on the human interpreting the data to “ensure data is accurate,” as AI tools cannot be totally relied upon to produce results that don’t require verification manually.

The ability of AI tools to deliver trustworthy results is a point of emphasis for Holdsworth as well. She said, “It would be hard for IPs to trust this technology at first to be 100% accurate,” thinking that she would “feel the need to go behind the AI and check its work.” While the promise that AI tools have shown in helping with infection prevention in many areas is significant, questions Holdsworth has regarding these tools include “how would validation of this technology work, how time intensive would this validation be,” and “do IPs have the time and energy to do this right now?”

Other concerns surrounding AI focus on the possibility of bias. For instance, an IP “may miss a pattern” even when the AI tool “stated there was one,” leading to confusion and further emphasizing the need for manual verification of what an AI tool tells a clinician. Cybersecurity also remains a concern with AI tools, as they “require using a third-party system” and granting that third party system access to “sensitive patient data,” Lamphier says. “Legal teams, risk teams, and cybersecurity / IT teams should be involved in the process with” the infection prevention department “to determine if these platforms are secure and what would happen if there were a data breach.” Having a unified approach across teams can prevent these from becoming issues in the first place, and putting plans in place in the event of a data breach or some other cybersecurity issue can mitigate the threat if they do eventually arrive.

As far as next steps for the future in the infection prevention space, Lamphier said,“More monitoring technology such as electronic hand hygiene technology that are meant to aid the IP workflow but not replace the IP is of potential interest. Aiding the workflow of an infection prevention department and not replacing the infection preventionist themselves is a vital refrain that came through when both experts were asked about potential next steps, underscoring the irreplaceable importance of having a human being involved in the process, even when advanced technologies like AI are being used.”

According to Holdsworth, certain AI advances “in the sterile processing and endoscopy space” relating to the “use of boroscopes and identifying what is inside a lumen” have been of particular interest to her. She believes that these burgeoning technologies will “be a huge help to not just IPs, but all reprocessing areas to ensure patient safety and the prevention of infections.”

Concerning other technologies not associated with AI, Lamphier provided a laundry list of exciting advancements in areas she believed our readers should be aware of, including “UV disinfection (including OR lights), electronic hand hygiene monitoring systems, antimicrobial textiles, technology that improves air quality, data mining technology, and platforms that organize HAI data.” Holdsworth believes that, as far as advancements on the horizon that staff should be trained with, virtual reality will be a “big move for the future.” She commented, “We are going to see virtual reality as an emerging strategy for education over the next [couple of] years.”

About the Author

Matt MacKenzie | Associate Editor

Matt is Associate Editor for Healthcare Purchasing News.