Dr. Martha Boeckenfeld is a technologist known for her insights on innovations that require human expertise. As noted on her LinkedIn page, she has held senior positions at Fortune 500 companies such as Axa and UBS, and served on the boards of BlackRock, UniCredit, and Generali.
Three years ago, she created a metaverse to pursue her passion for digital transformation. Additionally, she co-authored the book "Fast Future Blurs," published by Wiley.
In a recent commentary titled "Your AI Radiologist Will Not Be With You Soon" on LinkedIn, Boeckenfeld argues that healthcare needs more human experts, not fewer.
Despite predictions from some commentators that artificial intelligence (AI) would eliminate radiology jobs, this has not been the case at the Mayo Clinic. Instead, the Mayo Clinic has hired more radiologists to collaborate with their AI systems.
Boeckenfeld references a test conducted by radiologist Rajesh Bhayana, which demonstrated how Google's Gemini 2.0 accurately diagnosed pancreatitis from a CT scan, emphasizing that the test was purely educational and not based on a real patient.
The Mayo Clinic's model shows that their AI can identify patterns of pancreatic cancer that are not visible to the human eye. While specialized medical AI can achieve higher accuracy for specific tasks, when combined with the insights and expertise of human radiologists, cancers can be detected up to 12 months earlier.
Dr. Rajesh Bhayana's study revealed that while AI may identify patterns, radiologists provide essential context. AI can flag thousands of potential issues, but every flagged item requires human verification.
False positives need further investigation, and patients require explanations from human experts.
As it stands, radiologists have a diagnostic accuracy rate of 61% overall, compared to AI's best performance at 39%. Nevertheless, AI acts as a second reader for every scan, identifying subtle patterns that humans might miss, maintaining consistency across high-volume screenings, and extending expertise to rural areas.
The concept known as the Multiplication Effect can be summarized as follows: one radiologist plus AI equals enhanced detection; 100 augmented departments lead to a new standard of care; and 10,000 AI-assisted readings result in earlier interventions.
Mayo Clinic recognized something that others overlooked, Boeckenfeld asserts: AI does not replace expertise; it increases its demand. Ultimately, at scale, radiology is transformed, not replaced.
Every finding from AI requires human judgment, and every diagnosis necessitates human communication.
Therefore, your AI radiologist isn't coming soon because AI without radiologists is merely expensive pattern matching.