AI may eventually help orthopedic surgeons decide who should undergo surgery and predict how patients will fare afterward.
Eric Cohen, MD, an adult reconstruction surgeon at University Orthopedics and director of The Miriam Total Joint Center in Providence, R.I., isn’t starting there. For now, some of the most tangible returns he sees are happening far from the operating room: answering phones, booking appointments, navigating prior authorizations and finding revenue that might otherwise slip through the cracks.
That distinction matters as orthopedic groups confront an expanding marketplace of AI products while reimbursement pressure makes every new technology investment harder to justify.
“I probably hear weekly from someone pitching a different AI product,” Dr. Cohen told Becker’s. “You have to ask yourself what differentiates that product from the others and what’s going to really make a difference.”
His answer, at least today, is less futuristic than much of the AI conversation. The best technology does not have to replace a physician or make a surgical decision. Sometimes, it just has to solve a problem that already exists.
AI’s unglamorous sweet spot
One of those problems at University Orthopedics is patient access. The practice operates a busy patient access center, but high call volume can create long waits. It has added an AI conversational agent that can schedule and change appointments through integration with the electronic medical record.
Dr. Cohen has watched patients appear on his own schedule through the system. He also got an unusually personal test of it when his mother needed to reschedule an appointment with a shoulder surgeon.
She opted to speak with the AI agent, canceled her existing appointment and asked to see the physician rather than a physician assistant. The system found another opening.
“She had a great experience,” Dr. Cohen said. The anecdote illustrates the role he believes AI is best equipped to play right now. The technology provides another path into the practice without eliminating the human one.
Complex situations can still be routed to staff, he said, and he does not believe practices should dismantle their patient access operations simply because an AI agent can answer some calls.
“That’s kind of where I see AI at this point, to augment what you already have,” he said.
The same principle extends deeper into the administrative side of the practice. Dr. Cohen pointed to AI-assisted prior authorization and denial letters, as well as tools that can help practices track delayed or denied claims and determine whether encounters are ultimately reimbursed at contracted rates.
With reimbursement pressure mounting, he sees particular value in giving practices another set of eyes on money they have already earned. That may be a less dramatic application than an algorithm assisting in surgery. For an orthopedic practice, however, it can be considerably easier to prove whether it works.
The AI test: Does it actually pay?
Dr. Cohen does not believe an impressive demonstration is enough to justify an investment.
His framework starts with a trial. Practices should establish what they expect a product to accomplish, test it within existing workflows and then measure whether it increases revenue, improves patient throughput or expands access. He also recommends getting feedback from the people who will actually use the technology, both before implementation and during the trial.
“If it’s not meeting at least two of those marks, then it’s probably not worth investing,” he said.
There is another question practices should ask: What does this product do that the growing number of competing products cannot? That question becomes more important when the cost of technology is weighed against reimbursement pressure.
“Spending more money on technology when reimbursements are lowering doesn’t always make sense,” Dr. Cohen said.
His approach is therefore deliberately incremental: Test first. Measure the result. Keep existing workflows running. Then determine whether AI has earned a larger role. “I would encourage people not to go all in on AI,” he said.
Where Dr. Cohen draws the line, for now
The closer AI gets to clinical decision-making, the more cautious Dr. Cohen becomes. He does not currently use AI directly for clinical decisions in his own practice. Some of his partners have experimented with clinical applications, he said, and the feedback has been mixed.
“You have to watch out for what it’s documenting and doing in some of the clinical decision making,” he said.
That raises a fundamentally different standard from scheduling an appointment or identifying a delayed claim. An administrative error can create friction. A faulty clinical recommendation can affect care.
Dr. Cohen also sees privacy, security and transparency as important considerations when AI enters the exam room. If it is involved in clinical decision-making, he believes patients should know and be engaged in that choice.
“Not everyone’s going to be comfortable with that in the clinical setting,” he said.
More broadly, Dr. Cohen said he has seen some of the focus shift. Where early attention centered heavily on office applications, he is now seeing greater use on the administrative and financial sides. The distinction is not that AI has no place in medicine. It is that, at this stage, Dr. Cohen sees it as an aid rather than an authority.
“Pairing AI with our own clinical decision-making is really the best approach,” he said. “Using it as an aid, that’s really how I think we should look at it at this point.”
The one thing AI still can’t replicate
There is another limitation Dr. Cohen believes gets lost amid predictions about AI replacing physicians: intuition.
An algorithm can identify patterns across vast amounts of information and flag imaging findings, medication issues or other details for review. What Dr. Cohen believes it cannot replicate is the judgment physicians develop through years of treating patients.
“There’s no way around human intuition, which AI lacks,” he said. That is where he believes some of the industry’s hype has moved ahead of reality, particularly around clinical and operating room decision-making.
“I think it’s a little overhyped that it’s going to replace the physician who has the clinical experience and intuition, which is really irreplaceable,” he said.
His benchmark is straightforward. “I would take the clinical opinion of a surgeon who’s been in practice for 30 years all day over AI,” Dr. Cohen said.
The technology will improve. He expects that. But he cautions against allowing enthusiasm for what machines can do to diminish the value of what experienced clinicians already know.
What could change the equation
Dr. Cohen’s caution about AI today does not translate into a modest view of its future.
Ask where he hopes the technology makes its biggest difference five years from now, and his answer moves directly into clinical care. He sees potential for AI to help surgeons assess indications for surgery and better predict postoperative outcomes before an operation takes place.
Surgeons already make those judgments using training and experience. AI could potentially add something medicine has struggled to provide: better quantification.
“My hope is that AI assists us with indications for surgery and predicting postoperative outcomes preoperatively to better inform patients,” he said.
The opportunity, in his view, starts with data. Healthcare already holds enormous amounts of information, but it is scattered across different places and often requires clinicians to assemble and interpret it. AI could eventually help consolidate those pieces into a more complete picture of an individual patient.
“That’s where I think the greatest gain can go,” Dr. Cohen said, “really give us the full clinical picture of someone and then help us make that decision.”
It is a considerably different vision from replacing the surgeon. For now, Dr. Cohen sees the equation more simply: AI is most valuable when it augments the people already delivering care rather than trying to replace them.
At the Becker’s 32nd Annual Meeting: The Business and Operations of ASCs, taking place October 29-31 in Chicago, ASC leaders, surgeons and healthcare executives will explore strategies to drive growth, enhance operational performance, navigate reimbursement challenges and prepare for the future of ambulatory surgery. Apply for complimentary registration now.
