Skip to main content
Attach a clinical study and a template of your EU MDR appraisal table, and Anara fills one complete table row using only what the paper actually reports. Every cell traces to a passage in the source. If the paper doesn’t report an outcome, the cell stays blank. For the clinical evaluation where every data point has to defend itself.

​
1. Describe the task

EU MDR requires a clinical evaluation appraisal table where each row is one study and each cell traces back to the paper. Nothing inferred, nothing pulled from memory, nothing generalized from similar studies. If a cell contains a value the paper didn’t state, the appraisal fails on review. Anara fills one row at a time using only the attached paper. The workflow is the same for every study: you provide the table format, Anara reads the paper, Anara fills the row, and any cell that can’t be filled from the paper stays blank. The constraint is the point. Here a medical device company’s clinical research scientist is writing the CER for a robotic cochlear implant navigation system and has 40 included papers to appraise against a fixed template.

​
2. Give Anara context

Required context
  • The clinical study paper as a PDF attachment or from your library.
  • A screenshot or description of your appraisal table format, including column headers. Anara matches your structure exactly.
Optional context
  • One or two rows you’ve already completed. Anara uses them to match your level of detail and your phrasing conventions.
  • The device or intervention being evaluated, so Anara can flag passages about comparator devices or adjacent indications that might otherwise read as relevant.

​
3. What Anara creates

A single completed table row, formatted to match your template, with every cell traceable to the paper. Below the row, Anara quotes the specific passages it used so you can verify each data point before the row goes into your CER. Cells left blank are flagged with a short reason (“not reported” vs “not applicable”). The row is the audit-ready unit: paste it in, confirm the quoted passages, move to the next paper.

​
4. Follow-up prompts

​
Write the discussion paragraph for this paper

Once the row is filled, turn it into the narrative discussion that accompanies the table entry.

​
Compare this paper to the last three in the library

When you want to flag where this paper agrees or contradicts papers you’ve already appraised.

​
Catch an inferred value

If a cell looks too well-filled, ask Anara to prove it.

​
5. Tricks, tips, and troubleshooting

​
How you word your prompt shapes what you get

The constraint “use only what this paper reports” is the core instruction. Without it Anara completes the row by filling blanks with reasonable-sounding clinical knowledge, which is exactly what EU MDR prohibits. Restate the constraint even if it feels redundant. Asking for the three source passages at the bottom turns every row into a self-auditing unit, which is faster than reviewing against the PDF later.

​
Check the output against your own understanding

Read the three source passages Anara quotes before you trust the row. A cell that looks well-filled but doesn’t appear in the quoted passages is the failure mode to catch. This happens most often on secondary outcomes and adverse event counts, where the paper reports the data in a table Anara may have interpreted differently than you would. When in doubt, ask Anara to point to the specific figure or table number in the paper.

​
What to do with the output next

Paste the row into your appraisal table, cross-check the three quoted passages, and move to the next paper. For a fixed-template CER the workflow holds across every paper in the bibliography, which is why it scales from a single study to several hundred. When the appraisal is complete, use the discussion paragraph follow-up to draft the narrative pass that accompanies the table.