Loading...
Disclaimer: Informational only — not medical advice. AI hypotheses are not clinically validated. Medic Oath is for informational and educational purposes only — not medical advice. AI-generated hypotheses have not been clinically validated. Always consult a qualified healthcare professional. Privacy · Terms
The mission
Thousands of diseases have no one searching for a treatment — not because they can't be solved, but because there's no profit in solving them. MedicOath searches anyway: it reads the world's biomedical evidence in every language, proposes existing drugs that could treat diseases they were never designed for, commits every prediction publicly and dated so its track record is verifiable — and keeps each discovery free for humanity.
Plenty of tools generate medical ideas. What makes this one worth trusting is how it holds itself accountable.
Every drug→disease prediction is timestamped and committed before the outcome is known. No cherry-picking — the hit rate over time is public and verifiable. Most tools only show you their wins; this one shows you the whole ledger.
It reads Chinese, Japanese, Indian, African and Latin-American literature and traditional medicine — not just English. That’s most of the world’s biomedical evidence, the part almost everyone else ignores.
CC0 public-domain data, AGPL code, no paywall, no patent. Bring your own AI key and the platform costs you nothing to use. The mission in three words: we took ours.
Every claim is checked against the papers it cites, screened against Cochrane reviews and retractions, and the strongest run a gauntlet of eight specialist agents. Contradictions are surfaced, never buried.
The platform looks large because a cure has many stages — but it is really one pipeline. Here is the whole path, start to finish.
Every disease is ranked by a Neglect Score — the gap between how much it harms people and how little research it attracts. The engine prioritises the diseases industry has no profit reason to pursue.
Papers (in every language), clinical-trial registries, traditional-medicine compounds, and a knowledge graph of diseases, genes, drugs and pathways are assembled into structured evidence — connections no one could spot reading papers one at a time.
AI proposes an existing, approved drug that could treat the disease — grounded in the evidence and scored across seven dimensions (confidence, mechanism, evidence convergence, novelty, tractability, impact, cost).
A learned ranker turns the graph into a timestamped drug→disease prediction, committed before the outcome is known. The registry is append-only and pre-registered, so the hit rate becomes a public, verifiable track record — not a cherry-picked highlight reel.
Each claim is checked against its cited papers, screened for novelty, cross-referenced against Cochrane reviews and retractions, and — for the strongest — run through an 8-agent deep investigation (safety gate, ethics, bioinformatics, conflict-of-interest). Contradictions are surfaced, not hidden.
Discoveries are timestamped on-chain as dated prior art, the whole dataset is CC0 public domain, and community funding pays only for the AI compute. The aim is simple: keep the discovery free for the people who need it.
A plain-language directory of every page — grouped by section, each with what it does and how to use it.
Explore a disease, read a prediction, download the entire dataset, or ask the assistant anything. Nothing is behind a paywall — this was built for humanity to use.