About Tilefish MD
Saltwater fish disease diagnosis powered by real hobbyist data and Bayesian stats. No guesswork, no fluff.
Our Data at a Glance
How It Works
Base Probabilities
Every disease starts with a base probability from real data. Ich makes up about 31% of cases, flukes 18%, bacterial infections 13%, velvet about 8%. These come from 7,900+ real forum threads, not guesses.
Symptom Matching
Each answer you give shifts the probabilities. "Yes" to white spots boosts ich. "No" to rapid breathing makes velvet less likely. "Unsure" counts as partial evidence so you don't have to be certain.
Species Risk
Tracks 7 fish groups with their own risk multipliers. Chromis are 4.2x more likely to get uronema. Tangs are 3.3x for HLLE. Clownfish are 2.1x for brooklynella. All from real data.
Timeline + Water Params
How fast things got bad matters a lot. Rapid decline points to velvet, brook, uronema. Slow onset points to HLLE, starvation. Water params like ammonia, salinity, pH, and temp also shift the results.
Expert Rules
11 combo bonuses and 8 exclusion rules on top of the stats. Rapid breathing + swimming into flow + rapid decline = velvet boost. No new fish in 2 months = parasites unlikely. Stuff the math alone might miss.
Smart Question Order
It picks the next question based on what would narrow things down the most. You usually get a solid answer in 8-12 questions instead of going through all 67.
Where the Data Comes From
Humblefish Reference Threads
Expert disease profiles covering ich, velvet, brook, uronema, flukes, HLLE, lympho, bacterial infections, and more. These give us the symptom weights, treatment protocols, fallow periods, and important warnings.
humble.fish →Community Threads
22.8k+ forum threads scraped, with 7.9k+ being actual disease discussions. From those we get disease frequencies, which species get what, symptom patterns, and treatment outcomes.
Scraped with rate limiting