Last summer, a machine watched 17 hours of surgical video, and then, without a human hand guiding it, removed a gallbladder. The procedure was a milestone, showing that AI could work into the operating room—and not just as an assistant. But it also was a distraction from the true power of AI to assist surgeons, which has little to do with cutting.
For most patients, a surgery does not begin with an incision or end when they are wheeled out of the operating room. It might start with a sore knee, a fall, or a radiology scan revealing bad news. Then there are the myriad choices one faces, from whether to have a surgery or not to picking the right surgeon and best procedure. And it continues long after the stitches come out, with physical therapy, scar tissue, and the frustrating timeline of healing. It is across this long spectrum of time—full of complex conversations and decisions—that AI has the power to transform surgery.
In many other corners of medicine, AI is already changing patient care. More than 80% of U.S. physicians say they use AI tools as a knowledge co-pilot to double-check diagnoses, help find drug doses, write clinical notes, read scans, and interpret test results. Emerging programs help doctors answer concrete “yes” or “no” questions, automate administrative work, and distill information.
Surgery, however, presents a different kind of problem.
More than 4 million people die within a month of surgery every year, making postoperative mortality a leading cause of death globally. Complications can unfold in the weeks before an operation, when a patient’s blood sugar, nutrition, and anxiety shape how well their body will cope with the procedure. Others can emerge during the long recovery afterward, when missed physical therapy appointments and unread discharge instructions slow healing. AI that could support patients and clinicians across that entire arc would be far more transformative for surgeons than any robot in the operating room.
Read More: Why It’s So Hard to Reach Your Doctor—and How to Actually Get a Response
We’re beginning to see AI tackling these challenges. Algorithms can now identify which patients are at highest risk of complications before they enter the operating room. Another AI program monitors how fast a hip or knee replacement patient is walking in the weeks after surgery and alerts their doctor if recovery is falling behind. More than 2 million robotic-assisted procedures are already performed in the U.S. each year, and AI-powered simulators are also transforming how surgical residents train.
AI could potentially help in dozens of other ways. At the top of our wish list: an AI tool to help patients pick the best surgeon and procedure for their specific condition and give unbiased advice on the pros and cons of each decision step. Conversational AI tools can help make surgery less scary by providing emotional support and personalized pre-operative and post-operative guidance. Surgery is long-term, physically strenuous, and often irreversible. Its outcomes depend on caregivers, physical therapists, case managers, home-health aides, and the patient’s own consistency—each of which could be supported by AI but isn’t yet.
Yet we must move forward carefully. A 2024 study showed that AI recommended different diagnostic tests for patients depending on their race and gender. In surgery, that bias affects more than who gets diagnosed. It shapes who is offered surgery at all, what kind of surgery is suggested, how much follow-up they receive, and whether they heal alone or with help.
There are other questions, too. When an AI program gets something wrong, who bears responsibility? AI can already monitor surgeon’s skills and performance during robotic procedures—should those data be made public? And as AI takes on more of the work of surgery, will the next generation of surgeons lose the skills that come only from experience?
The engineers and health systems building new surgical AI tools should resist the urge to focus only on the obvious—the robot arm that can carry out a procedure autonomously, for instance. The equally important work happens in the days or weeks before a patient reaches the operating room and the months after they leave it. Preventing even a third of today’s surgical complications would save millions of lives and tens of billions of dollars.
The next great leap in surgical AI might not make headlines the way a robot removing a gallbladder does. Instead, it could look like a patient catching a surgical complication before it becomes a crisis or finishing physical therapy because an app noticed she was falling behind. Multiplied across 300 million surgeries a year worldwide, that’s where the potential to save lives lies.
The post The Unexpected Ways AI Can Transform Surgery appeared first on TIME.
