Medical Diagnosis Requires a Human Touch

There is perhaps no medical information more important than an accurate diagnosis. It is not only an explanation of symptoms but also the entry point to care. It’s the basis for a treatment plan and the mechanism by which insurance providers grant coverage. 

And yet for millions of people, getting a correct diagnosis is a long and uncertain journey. In 2023 a study led by researchers at the Johns Hopkins University School of Medicine estimated that some 795,000 Americans become permanently disabled or die each year because dangerous diseases like infections and cancers are misdiagnosed. 

It might seem natural to turn to technical fixes, such as more accurate tests or more artificial intelligence. But better diagnosis won’t come from technology alone; it will also require a more human approach. 

Brian Garibaldi, director of the Feinberg School of Medicine’s Center for Bedside Medicine, embodies this balance. Since launching the center in 2024, he has sought to figure out the best ways to combine emerging diagnostic technologies with traditional clinical skills to improve patient outcomes. 

“Over the past 30 years, maybe longer, there’s been a decline in overall bedside skill,” he told me. At the center’s inaugural conference in November 2025, young doctors and veteran faculty participated in hands-on examination sessions with real patients to help build their physical exam skills. “People who have been in practice for decades were coming out and saying, ‘That’s the first time I’ve ever felt a spleen,’ or ‘that’s the first time I felt a kidney,’” recalls Garibaldi, who is the Charles Horace Mayo Professor of Medicine. 

When meeting a patient for the first time, Garibaldi will sometimes begin the visit in the waiting room and walk the patient to the exam room himself. He notes the strength of the patient’s handshake and observes their gait and breathing as they walk. In that short period of time, he has done “part of a cardiac exam, part of a pulmonary exam, part of a neurologic exam,” Garibaldi says. With these early impressions, he can begin piecing together a diagnosis even before the appointment officially begins. 

A focus on these skills isn’t a rejection of technology. Garibaldi embraces tools such as AI and ultrasound, seeing them as potential bridges back to the bedside rather than replacements for physical exam skills. “If in the future there’s a digital stethoscope with an embedded AI algorithm that’s better at diagnosing heart failure than I am, that would be amazing,” he says. “But there will still be a role for empathy and curiosity, and for the power of touch, as part of the healing process.” 

Those qualities may improve not only diagnosis but also trust, which is increasingly fragile in American medicine. A 2024 Gallup poll found that only 44% of surveyed Americans said the quality of U.S. health care was excellent or good — the lowest rating since 2001. Garibaldi believes doctors also prefer it when more of their time is spent in bedside care rather than on the administrative tasks that increasingly dominate medical care. 

“People rate their physicians not based on their actual outcomes but on how the physician makes them feel,” he says. Improving diagnosis through better doctor-patient interactions isn’t just practical — it can restore a measure of humanity and trust to a system that needs it. 

Alexandra Sifferlin ’12 leads health and science coverage for The New York Times opinion desk. She wrote The Elusive Body: Patients, Doctors and the Diagnosis Crisis, from which this essay is partially adapted.