Medical Diagnostician's Approach and AI's Emerging Role in Clinical Reasoning

Gurpreet Dhaliwal, a professor of medicine at UC San Francisco, demonstrated diagnostic reasoning at the Society to Improve Diagnosis in Medicine's 2022 conference in Minneapolis. During a presentation, he correctly identified a patient's condition as abdominal pressure buildup that could lead to organ failure. Dhaliwal is recognized for his diagnostic skills and his ability to make clinical thinking transparent for educational purposes. He rejects the notion of being a 'master diagnostician,' viewing diagnostic excellence as a continual learning process. Diagnostic errors remain a significant concern in medicine, with a 2023 study estimating 371,000 annual deaths and 424,000 disabilities following misdiagnoses. Research indicates many errors stem from cognitive biases and judgment failures. Dhaliwal attributes his abilities to careful self-observation and tracking his cases to learn from mistakes. He follows William Osler's principle of rigorous self-assessment without self-deception. Recent developments involve artificial intelligence in diagnosis. Dhaliwal participated in a Clinicopathological Conference competing against an AI tool called Dr. CaBot, developed by Harvard Medical School researchers. Both reached correct diagnoses, though Dhaliwal identified a specific cause (swallowed toothpick) based on prior experience. Studies show AI models like GPT-4 achieving high diagnostic accuracy, sometimes outperforming physicians. Experts note AI's potential to assist with routine cases and complex, rare conditions, possibly reducing diagnostic disparities. However, concerns about reliability and the risk of skill erosion exist. Dhaliwal believes AI will transform healthcare but not fundamentally change doctoring, particularly in the 'muddy middle' of medical practice where value-based decisions require human judgment. The essay was adapted from Alexandra Sifferlin's book 'The Elusive Body: Patients, Doctors, and the Diagnosis Crisis.'