Attending to instincts  

By Anita Gupta, DO, PharmD, MPP, Johns Hopkins Medicine

The operation went smoothly and the patient tolerated the procedure with no problems. The surgical team turned the patient over to me in the post-anesthesia care unit (PACU) to wake up. Typically, that’s where the story gets boring. Not today.

I was the anesthesia provider responsible for caring for the patient throughout the perioperative period and stopped by to see how they were doing. Everything looked fine at first. The monitors showed reasonable numbers. No alarms were blaring. No flashing red indicators. All lab values were within normal limits. But the patient said they didn’t feel right, looked pale, and seemed more uncomfortable than usual following surgery. But blood pressure was low-normal, and the heart rate was slowly creeping up. Individually, none of these factors would have concerned me. But cumulatively, they pointed toward the dreaded diagnosis that every perioperative physician hopes to never encounter—internal bleeding. Subsequent evaluation bore out what I suspected. Soon, the patient was back in the operating room, where surgeons located and controlled bleeding. The patient ultimately recovered—but that’s not usually how these situations end.

Major bleeding after a smooth surgical procedure is rare. Even rarer are those instances where you know something is wrong before blood tests, imaging studies, or monitoring systems have clearly documented what’s occurring. This is a moment that reminds us that medicine—for all its advancements in technology—is a human-centric profession. But over the past few years, that statement sounds less true. Every day we hear about AI being used to extract information from medical records, interpret EKGs and radiology studies, predict patient complications, and support clinical decision-making. Much of this is exciting and some of it will improve patient care and of course we need to embrace it.

But when I recognized that something was wrong with the patient in the PACU, I didn’t review a vital sign that fell outside of normal parameters. There was no popup alert on my screen. No AI algorithm detected and notified me of an impending crisis. What happened was far less quantifiable. My concern was based on a gestalt assessment of the patient. Appearance. Movement. Surgical history. Decades of experience caring for patients recovering from anesthesia. Clinical instinct. All of this together can’t be fully replicated by AI.

Sometimes, experienced healthcare professionals catch problems before monitoring systems and computers do. The future of medicine will doubtlessly incorporate AI into every aspect of healthcare. And it should. But AI can only identify patterns. What happens in that recovery room is a demonstration of what separates humans from machines. One collects data. The other judges the nuances of human life. The future of medicine will be effective when it seamlessly combines both. Because when a patient is quietly bleeding to death in a recovery room, the difference between life and death may start with the thing no algorithm is programmed to do—the seasoned clinician at the bedside knowing something is wrong before the monitors catch it. Medicine may be heading toward a digital future. But some of healthcare’s most important decisions will always be human.

This piece expresses the views solely of the author. It does not necessarily represent the views of any organization, including Johns Hopkins Medicine.