Skip to main content
Turn a published clinical or scientific paper into a structured, verified summary ready for downstream medical writing. This is the starting point for most evidence-based workflows — from key message extraction to content outlines and briefing documents.
Risk tier: Low  ·  Review requirement: Standard medical writing review

​
What AI does

  • Extracts and organises the standard elements of a paper (design, population, endpoints, results, conclusions) in 2–3 minutes
  • Produces a working draft that frees you to focus on verification, emphasis, and contextualisation
  • Applies consistent structure across papers with different reporting styles

​
What AI cannot do

AI can misrepresent numbers, swap comparators, or conflate study arms. It does not know what matters most for your project. Every data point and conclusion must be verified by you against the original paper.
Do not use an AI-generated summary as source material for promotional or regulatory documents without full human review and verification against the original paper.

​
Before you start

  • Read the abstract, results, and conclusions yourself before generating a summary. You need to understand what the paper says before you can evaluate an AI summary of it.
  • Have the full text of the paper available — do not rely on the AI’s training knowledge of a paper. Provide the actual text.
  • Know your target summary format and any focus areas (e.g., “primary endpoint only,” “safety results”).

​
Steps

1

Read the paper yourself

At minimum, read the abstract, results, and conclusions before running this workflow. You need enough familiarity with the paper to catch errors in the AI output.
2

Prepare your inputs

Gather the full paper text and decide on your output requirements: summary format (structured abstract, narrative, bullet points), target length, and any focus areas.
3

Run the summary prompt

Provide the full paper text to your AI tool using the prompt pattern below. Specify the output structure and any focus areas in your prompt.
4

Verify every data point

Open the source paper’s results tables and verify every numerical value — p-values, confidence intervals, hazard ratios, percentages, and sample sizes — side by side with the AI output.
5

Check study arms and populations

Confirm that each result is attributed to the correct arm (treatment vs. comparator), population (ITT, mITT, per-protocol), and analysis type. This is where merging errors are most common.
6

Verify safety coverage

Cross-check that safety findings from the paper’s safety section are present and proportionate in the summary. An efficacy-forward summary that minimises AE data creates problems for every downstream deliverable.
7

Review conclusions

Compare every conclusion statement in the summary against the authors’ own Discussion and Conclusions sections. Flag any statement that overstates what the authors say.
8

Correct, refine, and finalise

Fix inaccuracies, fill gaps, adjust emphasis for your project context, and confirm the summary meets your length and format requirements.

​
Prompt pattern

You are a medical writing assistant. Your task is to summarise the following published paper into a structured summary.

Structure your summary with these sections:
- Citation (authors, journal, year)
- Study design and objective
- Population (key inclusion criteria, sample size)
- Primary endpoint and results
- Key secondary endpoints and results
- Safety findings
- Authors' conclusions
- Limitations noted by the authors

Rules:
- Base your summary only on the content of the provided paper. Do not add information from other sources.
- Reproduce data points (p-values, confidence intervals, hazard ratios, percentages) exactly as stated in the paper.
- If a finding is from a subgroup or post-hoc analysis, state this explicitly.
- Do not interpret the results beyond what the authors state.
- If you are uncertain about any data point, flag it with [VERIFY].

Paper text:
[INSERT FULL TEXT]

​
Human review checklist

Work through this checklist before using the summary in any downstream deliverable.
  • All data points (p-values, CIs, HRs, ORs, percentages, sample sizes) match the source paper
  • Study design is correctly described
  • Population and key inclusion/exclusion criteria are accurate
  • Primary endpoint result is correct and attributed to the right analysis (ITT, mITT, PP)
  • Secondary endpoints are accurately summarised
  • Safety data is present and not minimised
  • Conclusions match the authors’ stated conclusions
  • No unsourced claims or AI-generated background information
  • Subgroup and post-hoc results are clearly labelled as such
  • Summary length and format meet the project requirements

​
Common failure modes

RiskWhat to look for
Incorrect data pointsTransposed hazard ratios, wrong p-values, swapped arm results — verify every number against the source table
Merged study armsDrug arm and comparator results, or ITT and per-protocol populations, combined into one statement
Omitted safety dataSafety findings absent or reduced to a single sentence — cross-check the paper’s safety section
Overstated conclusionsSummary says “significantly improved” for a secondary endpoint or a non-significant trend
Hallucinated contextBackground sentences about disease prevalence or standard of care not sourced from the paper

​
Relevant tools

PosterLens

If your source is a scientific poster rather than a paper, use PosterLens to extract structured content first.

Source analysis prompts

Ready-to-use prompt templates for summarising different paper types.

​
Next steps

Extract Key Messages

Identify the evidence-supported messages from your summary.

Build a Content Outline

Structure a deliverable from the summary and key messages.

Final Review

QC checklist before the summary is used in a deliverable.