About Zebra Scout
Why this exists
In 2023, a mother named Alex used ChatGPT to identify her son's condition — tethered cord syndrome — after 17 doctors over 3 years had failed to connect the dots. She didn't use AI to replace her doctors. She used it to ask better questions.
This is the diagnostic odyssey. On average, rare disease patients see 7.3 physicians over 4.8 years before getting a correct diagnosis. For some, it takes decades.
Zebra Scout exists because no parent should have to become a medical detective — but until the healthcare system catches up, tools that help people ask better questions can save years of suffering.
How it works
Zebra Scout uses a curated database of 65 rare conditions and 241 symptoms, mapped to the Human Phenotype Ontology (HPO) — the standard medical vocabulary for describing clinical features, maintained by the Jackson Laboratory.
When you enter symptoms, the tool:
- Searches 241 HPO terms (plus synonyms) to find standardised matches for your symptoms
- Compares your selected symptoms against the known symptom profile of each of the 65 conditions in the database
- Calculates a symptom overlap score — pure set mathematics measuring how much your symptoms overlap with each condition's profile (40% Jaccard similarity + 60% condition coverage)
- Returns up to 15 conditions ranked by overlap score
This is not AI and not probabilistic. There is no language model, no neural network, no machine learning. The same symptoms will always produce the same results. The score is a measure of symptom overlap, not a probability of having the condition.
All computation happens in your browser. No data is sent to any server. No accounts needed. You can verify any result by following the Orphanet link on each condition.
What this is NOT
- Not a diagnosis. Only a qualified medical professional can diagnose a condition. A high overlap score means your symptoms appear in the condition's profile — it does not mean you have it.
- Not comprehensive. The database covers 65 conditions and 241 symptoms. The full Orphanet catalogue has over 6,000 rare diseases, and the HPO has 17,000+ terms. A condition not appearing in results may simply not be in our dataset.
- Not a replacement for medical care. Use this to start a conversation with your doctor, not to end one. Bring the results page to your appointment as a starting point for discussion.
The data
Conditions and symptom mappings are curated from public sources including Orphanet, OMIM, and the HPO. This is a prototype — a production version would integrate these databases directly via their APIs.
Why “Zebra”?
There's a saying in medicine: “When you hear hoofbeats, think horses, not zebras.” It teaches doctors to consider common diagnoses first. But for rare disease patients, the answer really is the zebra — and that saying can become a barrier to diagnosis. The zebra stripe ribbon has become the symbol of the rare disease community.
Open source
Zebra Scout is part of the willworth monorepo. The data and matching algorithm are open source — contributions welcome.