Hospital and health system leaders see significant potential for using artificial intelligence (AI) to transform activities ranging from treatment plans to diagnostics, administrative tasks and more.
During a Becker's Healthcare webinar sponsored by Elsevier, four healthcare experts discussed how their organizations are integrating AI into clinical decision-making and shared lessons learned:
- Abigail Baldwin-Medsker, MSN, RN, senior director, emerging digital programs, digital informatics & technology solutions, Memorial Sloan Kettering Cancer Center (New York)
- Paul Helmuth, MD, physician clinical executive, Elsevier
- Morgan Jeffries, MD, associate medical director for AI, Geisinger (Danville, Pa.)
- Maulin Shah, MD, chief medical information officer, Providence (Renton, Wash.)
Four key takeaways were:
1. Don't deploy AI just for the sake of deploying AI. It's essential for hospitals and health systems to use AI to solve specific problems and to drive tangible value. "We need to use AI where it makes sense," Dr. Shah said. "We have a whole repertoire of tools to influence behavior. I need an AI council to incorporate new technology in the right way, but I don't want that group to be a place where we say, 'Let's AI all the things.'"
Instead, AI solutions must be integrated into physician and researcher workflows in ways that ensure safe and effective care delivery. "We don't just want to implement AI to implement AI. We need to be thoughtful about the ways we are utilizing it as a solution," Ms. Baldwin-Medsker said.
2. Leverage AI so that employees can return to the more human elements of their work. Clinicians spend a lot of time in patient encounters, and they also spend a lot of time outside of those encounters on documentation. "The more we can use technology to free them up to be having a good face-to-face encounter with the patient, the more we will empower providers to make good decisions and keep them engaged in their work," Dr. Jeffries said.
Providence is taking a similar approach. "Generative AI and the automation part is really about taking away the drudgery and leaving clinicians free to do the stuff they want to do and to take care of humans," Dr. Shah said.
3. Avoid implementing too many AI point solutions. From a governance perspective, some organizations feel that a platform approach to AI is preferable to using multiple point solutions.
Providence focuses its technology efforts — including AI — in one area, such as patient message management. "We want to bring a lot of things to bear in that space, so clinicians feel like it's a real improvement," Dr. Shah said. Geisinger prefers enterprise solutions for AI. The organization is piloting an AI scribe product and recently made Microsoft Copilot available to clinicians.
4. Consider using AI to develop an entire ecosystem that supports clinical care. Many healthcare organizations are excited about using AI to document and summarize information from the electronic health record. But, they are thinking more broadly than just using AI for documentation.
"Is there a way to look at patient data and maybe social risk data from public sources, compare that to a knowledge database and deliver insight?" Dr. Helmuth asked. "If we can solve the safety and privacy concerns, it may be possible to care for patient populations more efficiently and bring forward more health equity."
To register for upcoming webinars, click here.