Rare diseases hide in plain sight. We help you find them sooner.
Rare Detect screens your symptoms against 8,000+ rare and orphan conditions using the same Human Phenotype Ontology trusted by clinicians worldwide. Free. Private. Built in India.
Rare isn’t actually rare.
Rare diseases are individually uncommon and collectively everywhere. The delay isn’t a failure of medicine — it’s a failure of pattern recognition at the point of first contact.
people will live with a rare disease at some point in their lifetime.
average time to a diagnosis, across visits to roughly 8 different doctors.
of rare diseases have no approved treatment — early identification is often the only intervention.
rare conditions. General practitioners aren’t trained to recognise patterns across all of them. They aren’t supposed to be.
A second opinion before your first appointment.
Rare Detect is not a diagnosis tool. It’s a screening companion.
Listens
Takes your symptoms in plain language — English or Hindi. No medical vocabulary required.
Translates
Converts what you wrote into formal medical terms (HPO codes), the way a specialist would.
Ranks
Orders the most likely rare conditions using deterministic Bayesian maths — not a black-box language-model guess.
Tells you what to ask
Gives you the questions to raise with a real doctor, and which specialty to seek out.
Flags emergencies
If your symptoms suggest something like a thunderclap headache or a stroke, you see an emergency warning before anything else.
Seven layers between your symptoms and our suggestions.
Most “medical AI” apps are a single language-model call behind a system prompt. Rare Detect is engineered like a clinical pipeline.
| Layer | What it does |
|---|---|
| 1. Parsing | Extracts symptoms from your free-text input. |
| 2. Crisis gate | Detects emergencies (stroke, thunderclap headache, sepsis) and fires an emergency card immediately. |
| 3. Screening | Filters out symptoms that don’t pattern-match any rare disease. |
| 4. HPO mapping | Maps each symptom to its standardised Human Phenotype Ontology code. |
| 5. Bayesian ranker | Pure deterministic maths — the same inputs always produce the same outputs. |
| 6. Policy engine | Decides whether to show a condition card, a “see a GP” nudge, or stay silent. |
| 7. Response generation | Composes a plain-language answer with its sources. |
Built for the person actually holding the phone.
Plain language input
Type symptoms the way you’d say them to a friend. No medical jargon required.
Photo of reports
Upload a lab report, prescription, or doctor’s note and we extract the relevant details.
HPO-grade matching
The same ontology used by Orphanet, the NIH, and the European Reference Networks for Rare Diseases.
Emergency-first
Life-threatening symptom patterns interrupt everything else with clear emergency guidance.
Privacy by design
Your symptoms never train a model. No data sold. No tracking pixels. Self-hosted backend.
Built for India
Bilingual (English and Hindi), runs on phones as old as Android 6.0, designed for unreliable mobile data.
We’re not another chatbot skin.
| Chatbot-style symptom apps | Rare Detect |
|---|---|
| Guesses based on language-model training data | Deterministic ranker over curated medical ontology |
| A different answer every time you ask | Same symptoms → same ranking, always |
| No way to audit why it said what it said | Every suggestion traces back to specific HPO codes |
| Trained on forum posts and consumer health sites | Built on Orphanet, ClinVar, and the Human Phenotype Ontology |
| “Talk to your doctor” disclaimer at the bottom | Refuses to answer when symptoms don’t match any pattern |
What Rare Detect will never do.
- Diagnose you. That’s a doctor’s job.
- Prescribe or suggest medication.
- Replace emergency services. We point you to them, fast.
- Sell your health data. We don’t have a business model that requires it.
- Pretend to be confident when it isn’t. If your symptoms don’t pattern-match, we say so.
Built on the standards your specialist already uses.
Human Phenotype Ontology
16,000+ standardised clinical features.
Orphanet
The European reference portal for rare diseases — 8,000+ entries.
ClinVar
The NIH database of genetic variants and their clinical significance.
GA4GH Phenopackets
The Global Alliance for Genomics and Health data standard.
Questions people ask us.
Yes. Permanently free — no ads, no premium tier, no in-app purchases.
No. You can use the app fully anonymously. Optional sign-in lets you save your history across devices.
Your symptom entries are processed on our backend to generate the response, then discarded. We do not train any model on user data, and we do not sell or share data with third parties. The full privacy policy is available in the app.
The app opens offline, but symptom analysis needs an internet connection because the matching engine runs on our backend.
Apple’s Developer Organization enrollment requires a D-U-N-S Number, which takes a few weeks to obtain. iOS is coming.
Yes. We’re piloting an enterprise version for hospitals and rare disease centres — get in touch.
Rare Detect is a wellness and information tool, not a medical device. It is not regulatory-approved and does not need to be, because it does not diagnose. It exists to help patients have better conversations with qualified doctors.
English and Hindi at launch, with more Indian languages on the roadmap.
GNH India — a team focused on closing the rare-disease diagnostic gap in India and emerging markets.
India has 70+ million people living with rare diseases. Most will spend years bouncing between general practitioners who, through no fault of their own, are not trained to recognise patterns across 8,000+ orphan conditions. We built Rare Detect because the diagnostic gap isn’t a doctor problem — it’s a tooling problem. And tools can be built. — The GNH India team