Casey Tingle , 2025-04-24 18:33:00
April 24, 2025
4 min read
Key takeaways:
- AI can be applied in clinical, administrative and research settings in health care.
- Surgeons must remain the decision makers in patient care and only use AI to assist.
SAN DIEGO — Speakers at an American Academy of Orthopaedic Surgeons Annual Meeting town hall said with the use of AI on the rise in many facets of daily life, one area of adoption that should not be overlooked is in health care.
According to the National Academy of Medicine, AI “offers unprecedented opportunities to improve patient and clinical team outcomes, reduce costs and impact population health.” Health care is also “increasingly digital in nature,” Siddarth Satish, MS, vice president of AI at Stryker, said in the symposium.
“In an operating room, every tool you are using has some form of user interface that is digital,” Satish said. “Many of these new technologies are starting to be connected either to some local network within your hospital or even to cloud servers.”
And orthopedics is no exception, according to Satish.
“Whether it is robotic-assisted surgical systems [or] navigation, there are a lot of these connection points,” Satish said.
Different applications
According to Cody C. Wyles, MD, an associate professor of orthopedic surgery and clinical anatomy the Mayo Clinic, AI can be broken down into different buckets, depending on the application. From a clinical aspect, Wyles said AI can be used for predictive modeling and creating a unique profile for each patient to identify the best treatment options, reduce complications and improve outcomes.
Cody C. Wyles
“We are able to use AI now to help us make a better intraoperative plan, especially when we pair this with complementary technologies, like robotics, augmented reality [and] computer navigation,” Wyles told Healio.
In addition, AI should be used to package information “into something that is actionable and understandable from the surgeon level,” according to Jason L. Dragoo, MD, chair of the devices, biologics and technology committee at the AAOS and professor at the University of Colorado and Steadman Hawkins Clinic Denver. He said using AI to write clinic notes is one possibility.
“A lot of information is given back and forth, and the computers can take a 30-minute conversation with the patient and be able to parse it down into what are the important parts of the conversation,” Dragoo told Healio. “AI can extract data from the patient’s medical record … that can be valuable to guide treatment.”
AI may also decrease administrative burden, allowing orthopedic surgeons to focus more on patients and less on paperwork, according to Wyles.
“We are already seeing evidence of that with automated responses for simple questions to patients, helping with insurance approvals [and] helping summarize large bodies of information with patients that come from referrals,” Wyles said.
When it comes to research, Wyles said AI can leverage available registry data “in a much more effective and efficient way, and build out new registries in a much more robust fashion with a lot more patient information through large language models.”
Challenges with AI
However, challenges exist in relation to AI in health care, one of which involves sharing data across hospitals. According to Wyles, research data are currently siloed across institutions, as many believe their data to be a “scarce, highly valuable resource.” But he said, “No hospital with its own data will be able to solve the problems that we need to solve.”
“Everybody is afraid they are going to give away the $10 billion source of data, but I do not think that is true in the vast majority of cases,” Wyles said. “We do need to be cognizant of where value is being created and compensate parties appropriately, but until we use things like federated learning and share data in a way that is HIPAA compliant, we are going to be stuck spinning our wheels like we have been for decades.”
Whether data are shared or not, Wyles said AI models need to be explainable and trustworthy. Currently, work is being done to explain how AI models come to a conclusion through a technique called uncertainty quantification, according to Wyles.
“If you use a technique called uncertainty quantification, you can know if the model is 85% certain or 55% certain,” Wyles said. “But, right now, if you have a binary output in your model, you are not sure if it is 50.1% certain or 99.9% certain. We need to focus on adopting models and putting the models into clinical practice that are explainable, have uncertainty quantification and are trustworthy.”
But the most important aspect of AI technology is that surgeons remain the main decision makers in patient care, according to Dragoo.
“[AI] cannot be a decision maker. It can assist in the process of making decisions and give you more information to make a better decision, and that is the key to how AI can safely be incorporated in how it fits into orthopedic practices,” Dragoo said.
References:
For more information:
Jason L. Dragoo, MD, wishes to be contacted through Rio Hobbs at rio.hobbs@cuanschutz.edu.
Cody C. Wyles, MD, can be reached at wyles.cody@mayo.edu.
Sources/Disclosures
Collapse
Healio Interviews
Disclosures:
Satish reports being an employee of Stryker. Wyles reports being a consultant for Johnson & Johnson MedTech. Dragoo reports no relevant financial disclosures.
