BLOG: Transforming medical diagnosis: emerging patterns of human-AI collaboration

By Luciana D’Adderio | Chancellor’s Fellow in Data Driven Innovation

When we think about artificial intelligence (AI) in healthcare, we often imagine robots replacing doctors. But the reality is far more nuanced and potentially more transformative. A new paper published in npj Digital Medicine co-authored by Luciana D'Adderio (Centre for Medical Informatics, Usher) and David Bates (Harvard Medical School, Harvard Chan School of Public Policy) reveals how AI is fundamentally changing the way doctors diagnose patients – and why this matters for all of us.

Consider the traditional way doctors make diagnoses: they examine the patient, gather information, and gradually build their understanding of what might be wrong. It's like solving a puzzle, piece by piece. But AI is flipping this process on its head.

In the traditional approach, doctors would examine a patient, order scans, and slowly build their case. With AI, something different happens: the moment a brain scan is taken, AI algorithms immediately analyse it and share potential diagnoses with the entire stroke team. Instead of working forward to a diagnosis, stroke doctors now start with AI's suggestion and work backward, verifying its accuracy through multiple steps.

This shift is crucial for several reasons. First, in time-critical conditions like stroke, every minute counts. When AI can instantly identify a blocked blood vessel in the brain and alert specialists, it could mean the difference between recovery and permanent disability or death. Second, this new approach helps ensure that multiple stroke experts can weigh in simultaneously, potentially catching things that might have been missed in a sequential process.

But perhaps most importantly, this transformation shows us something interesting about the future of medicine. Rather than AI replacing clinicians, we're seeing the emergence of a new kind of medical practice – one where AI and human expertise work in tandem. Doctors aren't simply accepting AI's diagnoses; they're developing new skills and practices to verify and validate AI's suggestions, combining the speed and pattern-recognition capabilities of machines with human judgment and experience.

This isn't just about stroke care. Similar changes are beginning to appear in lung cancer screening, emergency care, and other areas where rapid, accurate diagnosis is crucial. It represents a fundamental shift in how medicine might be practiced in the future.

The message is clear: AI isn't just another tool in the doctor's bag – it's catalysing a transformation in how we approach medical diagnosis. Understanding this change is crucial because it affects everyone who might one day need medical care (which is to say, all of us). As we move forward, the challenge won't be about whether to adopt AI in healthcare, but how to do so in ways that enhance rather than diminish the human elements of medicine.

The future of healthcare won't be about choosing between human doctors and AI – it will be about finding the best ways to combine both, leveraging their respective strengths to provide better care for patients. And that's something worth paying attention to.

Cite as

D’Adderio, L., Bates, D.W. Transforming diagnosis through artificial intelligence. npj Digit. Med. 8, 54 (2025). https://doi.org/10.1038/s41746-025-01460-1