> Source: https://github.com/aipoch/medical-research-skills

Lay Summary for Cross-Disciplinary Teams

Converts technical research into a structured summary that clinical, wet-lab,

bioinformatics, product, and management teams can rapidly read and act on.

Position in the Research Pipeline

This skill sits midstream:

• Upstream (should exist first): Clear research question, defined objectives,

structured results, result narrative

• This skill: Translates that clarified content for non-specialist readers
• Downstream (natural next steps): Slide Deck for Lab Meeting, Graphical

Abstract Generator, Reviewer Response Drafter

If the user's research content is still vague or unstructured, prompt them to

clarify objectives and key findings first. A lay summary built on unclear input

will sound smooth but be factually imprecise — worse than no summary.

Step 1 — Gather Input

Ask the user to provide any of:

• Abstract, introduction, or results section
• Key findings in their own words
• A study summary or internal report

Also ask: Who is the primary audience?

• mixed (default) — all teams listed
• clinical — clinicians, medical staff
• wet-lab — bench scientists, experimentalists
• bioinformatics — computational scientists, data analysts
• product — product managers, translational teams
• management — leadership, funders, executives

If unspecified, use mixed and include all relevant audience bullets.

Step 2 — Extract Core Structure

Before writing, internally map the input to these five elements:

If any element is missing from the input, note it in the output and invite the

user to fill in the gap.

Step 3 — Write the Lay Summary

Use the output template in assets/output-template.md.

Writing principles:

• No unexplained acronyms — define on first use or remove
• Evidence boundary must be explicit: distinguish finding from interpretation
• Each audience bullet should be actionable, not just descriptive
• Quantify findings where possible ("3-fold higher", "in 4 of 6 subtypes")
• The summary must stand alone without access to the original paper

For audience-specific language guidance, read references/audience-guide.md.

Step 4 — Quality Check

Before delivering output, verify:

• [ ] No naked jargon or undefined acronyms
• [ ] Finding is accurate — not overstated, not undersold
• [ ] Evidence boundary is clearly hedged
• [ ] Each audience bullet is actionable
• [ ] Summary reads cleanly to someone with no domain knowledge

If a check fails, revise before presenting.

References

• assets/output-template.md — the standard 6-section output template with example
• references/audience-guide.md — language and framing guidance per audience type