Medical Literature AI Summarizer & Extractor
Turn PubMed/Embase abstracts and clinical papers into structured evidence: study design, sample size, interventions, and primary outcomes with P-values and confidence intervals.
N, P, CI
auto-extracted
upload supported
RCT
design detection
Structured extraction
Ready when you are.
- Study design
- —
- Sample size (N)
- —
- Population
- —
- Interventions
- —
- Comparator
- —
- Primary outcome (effect · P · 95% CI)
- —
- Secondary outcomes
- —
- Limitations / risk of bias
- —
Plain-language summary
Example: clinical trial comparison table
Extract several papers and line them up. Illustrative example only.
| Trial | Design | N | Intervention | Primary outcome |
|---|---|---|---|---|
| Trial A (2024) | RCT, double-blind | 1,204 | Drug X 50 mg/day | −18% events (P=0.003, 95% CI 0.71–0.92) |
| Trial B (2023) | Retrospective cohort | 486 | Drug X + standard care | HR 0.84 (P=0.04, 95% CI 0.70–0.99) |
Built for evidence-based reading
Extraction is generated by AI models (DeepSeek or GPT) — always verify against the source paper.
Clinical fields, not just prose
Pulls out study design, N per arm, interventions, and endpoints instead of a generic summary.
Keeps the statistics
Surfaces effect estimates, P-values, and 95% confidence intervals exactly as reported — no invented numbers.
PubMed / Embase ready
Works with exported abstracts and pasted sections, or a PDF you upload straight from the browser.
How it works
From a raw abstract to a structured evidence card in seconds.
01
Paste or upload
Paste an abstract/section, or drop a paper PDF — text is extracted in your browser.
02
Extract clinical data
The model detects design, sample size, interventions, and outcomes with statistics.
03
Copy into your notes
Copy or download the structured fields for reviews, meta-analysis tables, or journal club.
FAQ
Does it replace reading the paper?
No. It is a triage and note-taking aid. Always verify every extracted number, endpoint, and limitation against the original article before citing.
Can it read a PDF?
Yes. PDF text is extracted locally in your browser and sent as text — scanned image-only PDFs without a text layer will not work.
Will it make up statistics?
It is instructed to report only P-values, confidence intervals, and effect sizes that appear in the source, and to mark missing fields as “Not reported”. Still, verify against the paper.