Duke Orthopaedics is using artificial intelligence (AI) to improve patient care from expedited referrals to ambient transcription to treatment suggestions. Orthopaedic trauma surgeon Christian A. Péan, MD, MS, serves as the executive director of AI and information technology innovation for Duke Orthopaedic Surgery. “As artificial intelligence is being used increasingly, we want to make sure that we’re using it in a way that prioritizes access and excellence first and foremost,” Péan says.
The university’s long history of research into AI and machine learning makes Duke Health uniquely suited to developing and using AI solutions. “We have a raft of expertise and experience in developing these systems,” says Duke Health orthopaedic spine surgeon Brett Rocos, MD. “With the cases we see, we can feed these models a wide range of information, making it incredibly clever.”
Learn how to refer a patient to Duke Orthopaedics.
Improved care coordination through referrals and access
The department is using AI to help expedite referrals and increase access. “We are really committed to expanding our digital front door,” says Péan. “We’re collaborating closely with care organization partners to use AI to more rapidly schedule patients, in order to help us document and find patients the right specialists at the right time.” He adds that AI also helps Duke providers communicate with referring physicians quickly and clearly.
“We encourage any and all providers to send patients to Duke,” Rocos adds. “No problem is too big or too small for Duke.”
AI assistance in the OR
Although implementation is still in the early stages, Rocos notes that the department is already using AI to assist decision making: “AI doesn’t replace the surgeon, but it can aid in surgical decision making. It’s like GPS in a car: the surgeon is still driving and choosing if these directions are the correct decision.”
This assistance helps while planning and performing surgery. “In spine, we’re using AI capabilities to make real-time decisions,” Rocos explains. “These systems can evaluate patients then and there while they’re asleep in front of us. AI systems use precise measurements to tell us when we’ve done enough to correct the patient’s spine. We can get the results we want without additional risk.”
Rocos reports improved outcomes and reduced procedure times among the benefits. “We’ve been able to do less surgery to reach the same goals, leading to shorter recoveries and shorter stays.”
Safeguards vital to implementation
Both physicians agree that governance and safeguards are vital to implementing AI in a clinical setting. “It’s still humans touching the patient and making the ethical decisions,” Rocos says. “We want to keep that human connection.”
Péan agrees that humans still need to take the lead: “I like to say that with AI, you don’t really ‘trust but verify.’ You verify and then trust, and only in that order.”
As AI is used across the medical field, Péan concludes, “[All physicians] have to be careful to balance our enthusiasm for using it to improve patient care with really thoughtful guardrails to make sure that we’re using it safely and responsibly. You need to have a really firm governance structure in place to vet any of these tools that we’re using, and we’re certainly doing that at Duke.”