NextConsensus
Which medical claims are becoming harder for institutions to ignore?
Medical evidence moves faster than the institutions that rely on it. The gap between the two is where your deadlines live.
You already know when a paper publishes. What no feed shows is the slower thing underneath — a claim surviving its ninth revision, shedding the qualifier that contained it, turning up on three pages that used to disagree. By the time a guideline confirms it, your window has closed.
For the teams working against dates that do not move.
The problem
You see the evidence. You don't see when it's becoming institutional reality.
A label change is a snapshot of something that had been moving for years. Reconstructing that movement by hand is days of work per claim, so most teams read the snapshot and infer the rest.
A concern that persists through nine revisions is a different object from one that appears and disappears in a week. Same alert, opposite meaning — only the trajectory separates them.
Becoming harder to ignore
Six signs a claim is moving from possible to consequential.
Not every revision matters. We surface claims showing one or more of these:
Gaining stronger support
A competitor's benefit claim gains two systematic reviews and a guideline citation in six months.
Becoming harder to dismiss
A safety claim about your drug survives nine revisions in a year, each one removing a qualifier.
Spreading across influential sources
A claim moves from the drug page to the condition and drug-class pages. Three independent sources now carry it.
Appearing in regulator or guideline language
Public sources start using language that mirrors FDA or NCCN wording, before any formal action.
Being contested
The same qualifier is removed, restored, and removed again across three editors. Nobody fights over language that carries no consequence.
Losing qualifiers
"May reduce" becomes "reduces," and the hedge never comes back — the claim now asserts more than it did.
Claims are ranked by how many signs they show at once and how close they sit to your portfolio. One sign is a curiosity. Four is a claim you should have an answer ready for. Where the record is too thin, we say so instead of ranking it.
Architecture
The method is trajectory reconstruction. The product is prioritization. Forecasting is the research program.
Refract reconstructs claim trajectories from public revision history. NextConsensus reads them and ranks what deserves your attention. Whether the same patterns also improve forecasts of regulatory, guideline, and coverage actions is an open research question, kept separate below.
Emergence, wording evolution, evidence attachment, persistence, spread across influential sources.
Claim intelligence — what you buy today.
Tests whether trajectory features — persistence, citation quality, velocity, propagation — forecast rung 5–7 transitions. Prospectively registered, not yet calibrated.
Authority-transition forecasting — a hypothesis under test.
Calibration, baseline outperformance, decision-useful lead time. Earned, not asserted.
Enterprise application — depends on demonstrated forecasting skill.
What you receive
Claim intelligence, not a document feed.
Every ranked claim arrives with the evidence behind its position — so you can disagree with the ranking on the merits rather than take it on faith.
Pair with Discovery to see what is emerging across your area.
What we threw out
We had a result. It didn't survive its own audit.
An earlier cycle looked like a finding: public revision activity appeared to run ahead of authority action. It was an artifact — our positive cases had complete revision histories and our negative cases truncated ones, so the model read a hole in our own data, not anything about medicine.
Ingested correctly, the signal disappeared — none of it held. That is why forecasting sits in the research program and not in the product. What survived the audit is the trajectory reconstruction itself, and that is the part you can buy.
Public record
Every trajectory is reproducible from the public record.
Each event has a verifiable source and a reproducible hash. You can recompute any trajectory yourself and check it against ours.