video1.0<iframe src="https://www.loom.com/embed/c797186b5528470bace8afe2b3f8c46c" frameborder="0" width="1954" height="1465" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>14651954Loomhttps://www.loom.com14651954https://cdn.loom.com/sessions/thumbnails/c797186b5528470bace8afe2b3f8c46c-4d2e336310bd1477.gif447.519In this video, I walk through how I've been using generative AI tools like Doximity to stress-test my own clinical reasoning — not to replace my thinking, but to catch what I might be missing after I've already done the work. I share a real de-identified case: a 70-year-old male presenting with recurrent falls, a new progressive headache worse with straining, visual changes, and cognitive decline. I walk through my original assessment and plan, then show exactly what happens when I paste the full H&P into Doximity using a structured attending-style prompt. The AI flagged chronic subdural hematoma (even without a witnessed head strike) and giant cell arteritis — diagnoses I hadn't fully considered. I ended up ordering an ESR/CRP, which came back negative, but the process helped me close gaps in my reasoning before the patient's plan was finalized. The key principle: reason first, let AI react second. These tools should be stress-testing your thinking, not doing it for you. This video is a companion to my Inefficiency Insights article in Healthcare Huddle. Full breakdown linked below. 📩 Read the article here: [https://www.healthcarehuddle.com/p/how-physicians-can-use-ai-to-catch-what-they-miss]