Oncology Failure Atlas

Evidence

Public source-linked evidence layers used to support OFA research navigation.

11,525 terminated and withdrawn oncology trial records · Public data only · Linked signals, not proof

Evidence overview

OFA evidence means linked public source context. It does not mean proof of causality, clinical conclusion, regulatory conclusion, investment signal, or treatment guidance.

Each OFA record is built from ClinicalTrials.gov and joined against linked public layers — PubMed, OpenFDA, and CrossRef — where the current OFA matching process resolves a link. RecurSignal™ is a deterministic recurrence signal computed on top of the dataset. The OFA Research Brief Generator sits downstream and drafts a human-reviewable brief from a filtered evidence pack.

Evidence Coverage Snapshot

AACT matched11,525 / 11,525
PubMed signals1,493 / 11,525
OpenFDA signals7,794 / 11,525
CrossRef DOI links1,033 / 11,525
RecurSignal11,525 / 11,525
Mechanism harmonization11,525 / 11,525
Any MDIP signal11,465 / 11,525

Coverage means OFA found a structured public-source signal for that layer. Missing signal does not prove absence of evidence. OFA uses compact source-linked evidence fields to support research navigation, not clinical, regulatory, investment, safety, or treatment conclusions.

Evidence layers

ClinicalTrials.gov source record

Primary source for trial metadata, status, phase, sponsor, intervention, indication, dates, and reported stop/discontinuation language. Every OFA record is identified by its public NCT ID.

PubMed linked signal

Direct NCT match or title/intervention similarity where applicable. A "NONE" or missing PubMed signal means no linked PubMed signal currently available in OFA — not that no publication exists.

OpenFDA linked signal

Exact or fuzzy drug-name matching where applicable. A "NO_MATCH" result means no linked OpenFDA signal currently available in OFA — not that no safety or regulatory evidence exists.

CrossRef DOI link

Supplementary DOI-to-NCT link layer. Absence of a CrossRef link does not mean no publication exists; it means OFA did not resolve a CrossRef link via the current matching process.

RecurSignal™ historical context

Deterministic historical recurrence signal across all OFA records, computed from MDIP™ category, trial phase, and classification confidence. Capped 5–95. Not a prediction, probability, forecast, clinical risk score, investment signal, or treatment guidance.

OFA Research Brief evidence pack

The structured snapshot sent to the Claude-assisted OFA Research Brief Generator. Built entirely from OFA-filtered analytics — no external sources are added at draft time. See AI Transparency for the full disclosure.

Evidence availability summary

LayerCoverage
Total OFA records11,525
ClinicalTrials.gov source records11,525 / 11,525
PubMed coverage rows (linked or NONE)11,525 / 11,525
OpenFDA coverage rows (linked or NO_MATCH)11,525 / 11,525
CrossRef positive DOI links1,070 rows · 1,003 NCT IDs · 1,062 unique DOIs
RecurSignal™ scores11,525 / 11,525

"Coverage" means a row was computed for the trial — not that a positive linked result was found.

Stop-reason language and MDIP

ClinicalTrials.gov stop-reason language — primarily the why_stopped field — is the primary input for MDIP (Mapped Discontinuation Intelligence Patterns) classification where available. Missing or ambiguous stop-reason language is explicitly tracked through dedicated unclear categories (UNCLEAR_NOT_SPECIFIED and UNCLEAR_WHY_STOPPED_REPORTED). Evidence layers support human review but do not prove causality or final root cause.

How to interpret missing links

No linked PubMed, OpenFDA, or CrossRef signal currently available in OFA does not mean no evidence exists. It means no linked signal was found by the current OFA matching process. Reasons can include missing metadata, absent NCT IDs in publications, investigational compound naming gaps, or drug-name normalization issues.

Evidence Pack — what's inside

The evidence pack sent to the OFA Research Brief Generator is a structured snapshot of the user's current OFA-filtered scope:

  • selected filters (indication, mechanism, phase, sponsor, MDIP™ category, RecurSignal band, etc.)
  • record counts within scope
  • top MDIP™ discontinuation-pattern distribution
  • top harmonized mechanisms (from trial_mechanism_harmonization.harmonized_mechanism_class)
  • phase distribution
  • linked-signal availability summary (PubMed, OpenFDA, CrossRef)
  • a small set of representative records
  • known limitations of the current scope
  • explicit human-review requirement

What evidence is not

  • not medical advice
  • not a clinical conclusion
  • not a regulatory conclusion
  • not investment advice or diligence
  • not a prediction or risk forecast
  • not proof of drug, asset, or program failure
  • not a complete literature review

Continue

Oncology Failure Atlas · Public data only · Linked signals, not proof · No clinical, regulatory, investment, or treatment advice.