Clinical Trial Literature Search

Problem It Solves

When preparing trial planning, protocol/CSR background, or published-safety checks, evidence is scattered across OpenAlex, Europe PMC, and preprints; fields are inconsistent, records duplicate, citations can be hard to verify, and published safety literature is easily mistaken for a FAERS signal. This skill turns published-evidence retrieval into a repeatable JSON/Excel/HTML workflow.

How It Works

• Multi-source public literature search: uses OpenAlex as the primary source, keeps Europe PMC on by default, and can enable --cochrane, --with-biorxiv, --with-arxiv, and other opt-in sources.
• Normalization and de-duplication: merges records by DOI or title while preserving provenance, PubMed/PMC ID, abstracts, OA full-text URLs, MeSH terms, keywords, funders, and other structured fields.
• Evidence packaging: outputs lit_report.xlsx, lit_report.html, and references.bib / references.ris; --prisma can generate a PRISMA funnel, and --rank relevance scores title/abstract relevance.
• Safety and citation verification: --safety is only a qualitative filter for published AE/PV literature, while --verify checks citations against DOI, Crossref, Europe PMC, and other sources to reduce hallucination.

Boundaries

• Reads public bibliographic APIs only; it does not fetch registry structured metadata or perform FAERS disproportionality analysis.
• Defaults to SAFE PREVIEW; network requests require explicit --run. OpenAlex keyless access is quota-limited, so a local .env key is recommended.
• Guideline content is a local pointer-only corpus rather than a live latest-guideline fetch; PROSPERO/NICE/TRIP may be skipped due to token or documentation limits.
• Output is for background reference only and is not a regulatory submission.