Your first extraction

Your first job is to get your included studies into a review, confirm the machine can read each one, and run extraction over the ones that are ready. This page walks that path start to finish. Nothing here spends a credit until you confirm a run.

Bring your studies in

Open Import from the Studies list. There are three ways in, one per tab:

• Reference file: the export from your reference manager. Zotero, Mendeley, EndNote, RefWorks, Covidence, and Rayyan all produce RIS, BibTeX, .nbib, or CSV, and this tab reads all of them. This is the path for a screened set: one file, your whole included list.
• PMID / DOI list: paste a list of PubMed IDs or DOIs when you have identifiers but no export file.
• Upload PDFs: drop the PDFs directly. Around 45% of RCTs have no open-access fulltext, so bring-your-own PDF is a first-class path. It is also how you rescue any study that came in without a fulltext.

Every successful add lands the study in your Studies list immediately.

Watch each study settle into a state

After import, each study carries a status chip that tells you exactly where it stands. TrialExtract fetches and parses the source, and the chip moves through the sequence: Added, then Fetching fulltext while it retrieves the paper, then Reading, then Ready once the paper is readable, its identity is confirmed, and its design is supported.

Some studies stop short of Ready, and the chip says why rather than dropping the row:

• No fulltext: TrialExtract could not retrieve the fulltext, so the Studies-list chip reads No fulltext — hover it for the reason the fetch found nothing. A study with no PMID or DOI to fetch by shows No fulltext · no PMID/DOI in the import results. Either way the row offers to take a PDF: upload it and the study re-reads.
• Needs identity: the study needs a confirmed PubMed ID before it can run.
• Unsupported design: the design is not one TrialExtract extracts, so there is nothing to pull.

Only a Ready study is selectable for a run. That is the definition to watch for: readable, confirmed identity, supported design.

Preview any study for free

Click a Ready study to open its preview. The preview is free and spends no credit. It shows what TrialExtract read: the paper’s synopsis, how readable the source is, and an inventory of the source blocks it will extract from. If TrialExtract is still reading, the page says Still reading this paper and updates when the inventory is ready.

The preview is the check before you spend. Read it to confirm the machine is looking at the right paper and can actually see the numbers you need.

Run extraction over the ready ones

Back on the Studies list, tick the Ready studies you want and start a run. The run sheet is the one place a cost ever appears. Each ticked study shows its own per-paper price, and the sheet sums them: it never charges a flat rate times a count. The cost card lists Studies, Credits required, Account balance, and Balance after run, so the total is in front of you before you commit.

Confirm binds that quote. If per-paper pricing changed between the moment you opened the sheet and the moment you confirmed, the run re-quotes to the new total and asks you to confirm again. You are never charged a total you did not see.

Confirmed studies become review rows, each carrying a receipt for every value the machine pulled. See The receipt for what that provenance holds. You can keep adding studies while this runs.