Evidence on clinical AI models, answered with citations
Ask about any clinical AI model, benchmark or validation study. Get a synthesized answer from 47,000+ curated papers, with a reference for every claim.
3 free searches, no sign-up. PubMed · Europe PMC · arXiv · medRxiv · PMC · updated weekly
Several large language models have been evaluated for radiology report summarization. They include general-purpose models like BART and T5, which reached state-of-the-art results on benchmarks such as MIMIC-CXR and OpenI, and domain-adapted variants like SciFive and ClinicalT51. More recent evaluations include open-source models such as OpenChat-3.5-7B, Starling-LM-7B and Meta-Llama-3-8B-Instruct2.
Answers you can check, not just read
Clinical Corpus searches only a curated clinical-AI literature base and writes a short synthesis from what it finds. Nothing comes from the open web.
- Every claim cited. Click a number to open the paper behind it.
- Reference lists in seconds. Ask for “50 articles on …” and get Vancouver, APA or Harvard.
- Current. New PubMed, arXiv and medRxiv papers are added every week.
Built for people who build, buy or evaluate clinical AI
Medical-AI product teams
See which models, datasets and benchmarks are being published in your area, and cite the evidence in product, regulatory and investor documents.
Clinicians and researchers
Check what has actually been validated for your specialty before trusting a tool, and build the reference list for your next paper.
Analysts, HTA and market research
Answer “what is the evidence for X?” in minutes, with every statement linked to its source.
Ask. Check. Cite.
Ask in plain language
Any question about clinical AI models, benchmarks, validation or deployment, in English or your own language (answers are in English).
Get a cited synthesis
A concise answer first. Want more? One click asks again with more evidence for a detailed answer.
Export the references
Formatted reference lists (Vancouver, APA, Harvard), or print any answer to PDF.
47,000+ papers, only on clinical AI
Language models, small and edge models, vision-language models and clinical prediction models, curated from five databases and updated every week. Next: bioRxiv and OpenAlex.
| Source | Articles |
|---|---|
| PubMed | 18,278 |
| Europe PMC | 10,763 |
| arXiv | 7,093 |
| PMC | 4,022 |
| medRxiv | 2,926 |
| Other sources | 3,474 |
| Total | 46,556 |
PubMed adds over 4,000 articles a day. No physician, developer or medical team can keep up alone.
Clinical Corpus was built by Dr. Alexandre Bandeira de Mello: over 40 years of medical practice, 29 of them in emergency medicine, and online since before the Web (he founded Brazil’s second internet-connected BBS in 1993).
Read the story →Start free. Upgrade when it earns its place.
Free trial
- Full access, detailed answers and reference lists
- No credit card
Frequently asked
What exactly is in the database?
47,000+ papers on clinical AI from PubMed, Europe PMC, arXiv, medRxiv and PMC, curated to clinical AI research and updated weekly.
How do I know an answer is right?
Every claim links to the paper it comes from, so you can open the reference and check. Answers only use papers in the database.
Is this medical advice?
No. Clinical Corpus is a research tool for professionals studying clinical AI. It is not for decisions about patient care.
Do I need an account?
No. You get 3 free searches a month without signing up. The free 7-day trial (no credit card) adds detailed answers and reference lists.
Can my team use it?
Yes. Team plans start at 5 seats with a manager dashboard, and custom seat counts are available.
What would you like to know about clinical AI?
Three free searches. No sign-up.