Guide
How to scan a research paper PDF into a structured summary
Any language, page-level citations, and the file never leaves your device.
Screening a paper properly means answering the same handful of questions every time: what kind of study is this, who was enrolled, what was compared against what, what was the primary endpoint, what did it actually report, and what did the authors admit it couldn't show. Doing that by hand across forty PDFs is where a review week disappears.
Symmathy's Scan a PDF feature does the mechanical part. You get a structured record of what the paper states, with the page number attached to every key finding so you can check it in seconds rather than trusting a summary.
What comes back
- Study type — RCT, cohort, meta-analysis, preprint, case report.
- Population, intervention, comparator, primary endpoint — the PICO frame, extracted verbatim where the paper allows it.
- Primary result — effect size, confidence interval, and p-value when reported.
- Key findings with page numbers — four to six results, each tied to the page it appears on.
- Limitations — as stated by the authors, not inferred.
- Funding, conflicts of interest, registration ID — the provenance fields reviewers keep having to hunt for.
What you don't get is a score. Symmathy does not grade evidence, rank papers, or recommend one study over another. Appraisal is the reviewer's job, and a tool that pretends otherwise is a liability in a methods section.
Where the PDF actually goes
Text extraction runs in your browser with pdf.js. The PDF binary is never uploaded. Only the extracted text is sent for analysis, and it passes through a patient-information scrub first, so identifiers that occasionally survive in case reports and appendices are removed before analysis.
Papers that aren't in English
The scan reads the paper natively — Japanese, Thai, Spanish, Portuguese, Chinese, French, Arabic, Korean — and writes the structured breakdown in the language you pick, while keeping the original title verbatim so it stays citable. It also proposes short English search phrases from the paper's own terminology, which is how you turn one regional trial into a wider federated search across all twelve databases.
Fitting it into a review workflow
- Run the question as a federated search and save the candidates.
- Pull the open-access full text where it exists, or upload the PDF you already have.
- Scan each paper and read the structured fields side by side.
- Send two or more papers to Compare when the endpoints line up.
- Keep notes on each record, and record counts and dates for your PRISMA flow.
Questions people ask
How do I summarize a research paper PDF without uploading it to a server?
Symmathy extracts the PDF text in your browser using pdf.js. The file itself never leaves your device — only the extracted text is sent for analysis, and it is scrubbed for possible patient identifiers first.
What does Symmathy extract from a research paper?
Study type, population, intervention, comparator, primary endpoint, primary result with effect size and confidence interval where reported, key findings with the page they appear on, limitations, funding, conflicts of interest, and any trial registration ID.
Can it read a paper written in Japanese, Thai, Spanish, or Portuguese?
Yes. The paper is read natively in its source language and the structured breakdown is written in your chosen output language, while the original title is kept verbatim.
Does Symmathy grade or rank the paper?
No. Symmathy reports what the paper states and cites the page. It does not assign quality scores, evidence grades, recommendations, or clinical advice. Appraisal stays with the reviewer.
Can I ask follow-up questions about the paper?
Yes. After a scan you can ask questions against the extracted text, and answers point back to the page in the PDF so you can verify them.