Methodology

How we source, process, and publish research findings.

Sources

CancerRadar pulls data exclusively from authoritative primary sources:

  • PubMed / NCBI E-utilities — peer-reviewed biomedical literature
  • ClinicalTrials.gov — registered clinical trials
  • Europe PMC — open-access biomedical literature
  • National Cancer Institute (NCI) — authoritative cancer information
  • Crossref / OpenAlex — bibliographic metadata

We do not use health news sites, blogs, press releases, or other secondary sources as the basis for factual claims.

How findings are created

A daily crawler fetches new records from approved sources. Each record is:

  1. Deduplicated using DOI, PMID, or NCT ID
  2. Classified by cancer type
  3. Summarized into plain language grounded in the source abstract
  4. Assigned an evidence stage based on study type
  5. Reviewed before being published

AI-assisted summarization is used to produce plain-language explanations. All AI output is grounded in the retrieved source text. AI is not used to invent findings, citations, or trial outcomes.

Evidence stages

Every finding is labeled with an evidence stage so readers can understand how established a finding is:

Preclinical

Lab or cell studies. No human data.

Animal Study

Results in animals only. May not apply to humans.

Phase 1 Trial

Early safety trial in humans. Small groups.

Phase 2 Trial

Efficacy trial. Larger but still limited.

Phase 3 Trial

Large-scale comparison trial.

Observational

Associations observed. Not causal proof.

Review / Meta-Analysis

Summary of existing evidence.

What we never publish

  • Claims without a traceable source
  • Findings that imply cures or guaranteed outcomes
  • Content that sounds like medical advice
  • Duplicate or near-duplicate records
  • Records where the cancer type cannot be reliably identified

Limitations

CancerRadar does not cover all published research. Coverage depends on what is indexed by our approved sources at the time of each crawl. There may be delays between publication and appearance on this site.

Plain-language summaries are simplifications. They may omit technical nuance that is significant for specialists. Always read the original source for complete information.