DiffyD is an experiment in better clinical reasoning.
It takes the kind of messy, unstructured case notes we all write and turns them into structured, transparent differential diagnoses.
It's built for clinicians who want to think better, not faster. AI shouldn't replace clinical judgment; it should make it easier to reason clearly, teach effectively, and keep improving how we practise medicine.
Why I Built It
I've always felt that diagnostic reasoning is both the best and least supported part of medicine. We talk about it, we teach it, but there aren't great tools that help us see or share how we actually think.
DiffyD started as a way to fix that - to take advantage of what large language models can do with free-text vignettes, and to structure that into something useful: a transparent, explainable differential list that clinicians can interrogate, refine, or disagree with.
It's not about outsourcing thinking. It's about scaffolding it.
How It Works
Write or paste your case.
A few sentences, a ward note, a detailed history - it doesn't matter. DiffyD is designed to cope with the messiness of real clinical language.
AI does the heavy lifting.
It pulls out the key clinical features, recognises patterns, and builds a structured differential diagnosis with evidence and suggested investigations.
You make sense of it.
You get a transparent output you can agree with, challenge, or learn from - and use it to sharpen your reasoning, not automate it.
The Vision
The goal isn't a "diagnosis engine." It's a reasoning tool - something that helps clinicians, students, and teams think in a more structured, teachable way.
If we can make clinical reasoning visible and shareable, we can improve the quality of medicine everywhere.
This is early work - part education, part research, part curiosity. But the ambition is simple: build better tools for how clinicians actually think.
Technical Preview
DiffyD is still in technical preview. It's for educational and research use only - please don't use it for real patient care.
Like all AI systems, it can get things wrong. Clinical judgment always comes first.
Get in Touch
If you're curious, have feedback, or want to experiment with how this could fit into real-world practice or education, I'd love to hear from you.