Publishing Sympify's Diagnostic-Equity Research

The research question

Diagnostic error is not evenly distributed. People who cannot name their symptom in clinical vocabulary get worse search results, worse triage, and worse outcomes. The paper asks whether a reverse-dictionary approach — going from a plain-language description back to the term — can close part of that gap, and what it takes for the system to stay usable rather than just impressive.

Three parts

• Reverse-dictionary algorithms — mapping descriptions people actually use onto clinical terms
• Patient-AI collaboration — keeping the person in the loop rather than handing down an answer
• Sustainable AI design — treating compute cost and maintainability as an equity issue, not an afterthought

Add the methods and evaluation summary in publication-safe terms

Note on scope

This page is about the research. The story of building Sympify — the team, the chapters, the events — lives on the Community & Impact page.