Living with a Rare Disease
There are approximately 7,000 rare diseases. Collectively, they affect 300 million people worldwide. This page exists because I'm one of them.
The Diagnostic Odyssey
I developed symptoms of a rare disease during my PhD. I lost the use of my hands and feet. I was misdiagnosed more times than I can count. My condition remains partially undiagnosed to this day.
The thing nobody tells you about a diagnostic odyssey is that you become the project manager of your own healthcare. You learn to carry your records between specialists, to explain your symptoms in the language each doctor expects, to advocate for tests that the system has no incentive to run. You do this while sick. While scared. While the bureaucracy treats you like a problem it wasn't designed to solve and moves on.
The data to help patients like me exists somewhere — a paper in a journal no one reads, a case study at an institution I'll never visit, a pattern in a dataset that no one has connected yet. The fragments don't talk to each other. And the patient — the person least equipped to coordinate a research program — becomes the only one trying to connect the dots.
That gap is not a medical failure. It is an infrastructure failure. And infrastructure is something I know how to build.
Why Patients Should Build the Tools
There is a specific kind of expertise that only comes from being inside the system. I know what it feels like when pre-existing conditions block access to quality care. I know what it means to sit in an ER at 3 AM writing symptoms in a journal because the chart doesn't capture what's actually happening. I know the weight of navigating fragmented medical records when you can barely hold a pen.
This is not a sympathy story. It is a design requirement. Systems must be built for the people who need them most, not optimized for the people easiest to serve. A patient's 3 AM symptom journal holds as much validity as a clinical observation. Both are data. Both deserve infrastructure.
The people closest to the problem should help build the solutions — not because patients are noble, but because we carry information the system can't see from the outside. When I walk up to a physician researcher and ask "How can I help?", the question carries specific knowledge. I know where the gaps are because I've fallen through them.
What I'm Working On
Lattice Protocol is federated research infrastructure so that compute is never rare for rare disease researchers. So that data moves across institutional boundaries while respecting privacy. So that the next patient navigating a diagnostic odyssey has tools built by someone who understands what that odyssey costs.
The Rare AI Archive is an open-weight collection of language models fine-tuned on rare disease literature — built with the Wilhelm Foundation so that any researcher can download and run these models locally, without gatekeepers or API costs. No disease is too rare to matter.
The Undiagnosed Patients Hackathon brings together clinicians, ML scientists, and patients in rapid-prototyping sprints. What started with Michael and me at Stanford has grown to six countries, 130 collaborators, and real diagnoses delivered to real patients in 48 hours.
These are not separate projects. They are the same project: closing the gap between what technology can do and what patients actually get. The Folding@Home kid grew up, got sick, and built the infrastructure he wished had existed.