← PICO-SEARCH

Sources

Every evidence source Nerd Burger and the 24 Sliders can reach, grouped by how they're accessed. This is the trust layer — no hidden content, no fabricated citations, no pretending we have what we don't.

Australian-first, globally complete. We prioritise Australian college guidelines and government health resources, then layer on the global colleges, WHO, NICE, and the full biomedical literature through PubMed, Europe PMC, OpenAlex, ClinicalTrials.gov and a growing list of specialist sources.

Legitimate sources only — no custom scraping

PICO-SEARCH only indexes content we have explicit permission to access. We use public APIs (PubMed, Europe PMC, ClinicalTrials.gov, OpenAlex, Semantic Scholar, CORE, Crossref), legitimate aggregators (Epistemonikos, TRIP, GIN International Library), and partner-based feeds (NICE Syndication API when approved). We never scrape custom sites, never cache content, and never paraphrase licensed material. Commercial references (eTG, AMH, UpToDate, DynaMed, BMJ Best Practice, Embase, CINAHL, MIMS, HealthPathways) are handled via respectful click-through only.

How we rank the evidence — CEBM pyramid

Not every study is equal. Every citation PICO-SEARCH returns is tagged by study type (from PubMed's publication_type when available, otherwise a regex over title + abstract) and mapped to the Oxford Centre for Evidence-Based Medicine (CEBM) hierarchy. Strong-top, weak-bottom. The question type (treatment / prognosis / diagnosis / aetiology) reshuffles the pyramid — for example, prognosis favours inception cohorts over RCTs.

1aSR of RCTs— Systematic review / meta-analysis of randomised controlled trials
1bSingle RCT— Well-conducted RCT with narrow confidence interval
2aSR of cohort— Systematic review of prospective cohort studies
2bSingle cohort— Individual cohort study or low-quality RCT (wide CI)
3aSR of case-control— Systematic review of case-control studies
3bSingle case-control— Individual case-control study
4Case series— Case series, case reports, poor-quality cohort / case-control
5Expert opinion— Narrative reviews, editorials, bench research, expert consensus
Treatment questions
Pyramid as shown. SRs of RCTs → RCTs → cohorts → case-control → case series → expert opinion.
Prognosis questions
Inception cohorts promoted above RCTs. A well-conducted cohort with long follow-up beats a short-follow-up RCT for "what's the natural history of X?".
Diagnosis questions
SRs of validating cohort studies > validating cohort > exploratory cohort. RCTs only matter if comparing two diagnostic strategies head-to-head.
Final ranking formula: rank_score = evidence_tier_weight × recency_decay × relevance_score × quality_multiplier. Tier weight: 1a = 1.00, 1b = 0.80, 2a = 0.60, 2b = 0.45, 3a = 0.30, 3b = 0.22, 4 = 0.15, 5 = 0.10. Recency decay: exp(−age_years / half_life) where half-life is 5y for treatment/diagnosis, 10y for prognosis. Quality multiplier: Jadad heuristic on RCTs (0.7 – 1.1).

Live APIs · 11

Queried on every search in parallel. Explicit public API with TDM rights or CC0 licence.

Local corpus · 3

Refreshed on a cron and stored on the VPS for millisecond lookups.

Partner APIs — live · 3

Accessed via an explicit partnership agreement or syndication API. Full attribution on every citation.

Free APIs — coming · 1

Client code ready, awaiting final wiring.

Partnership outreach pending · 72

Listed so users know we're aware of them. We're approaching each publisher with a polite partnership request for indexing rights — see docs/partnership-outreach-template.md. Never scraped without explicit permission.

Licensed — link-out only · 14

Commercial content we respect and never scrape. The Head Nerd surfaces a click-through when a question is in their territory.

Why we don't include eTG, AMH, UpToDate, Embase, CINAHL

Those products are commercially licensed and their publishers do not offer redistribution rights. PICO-SEARCH respects that — we never scrape them and we never fabricate their content. Instead, when the Head Nerd detects a question that is authoritatively eTG / AMH / UpToDate territory (dosing, drug interactions, PBS criteria, point-of-care decision support), the answer surfaces a "click through to [eTG / AMH / UpToDate] for the authoritative guidance" banner so the clinician opens the right source in their own subscription.

If your institution licenses one of those products and wants PICO-SEARCH to deep-link into your subscription, get in touch: ops@picosearch.ai.