gistfile1.txt

---
name: movement-science-lecture-update
description: Use this skill when updating a lecture with recent PubMed research. It extracts lecture objectives and baseline concepts from a slide deck, searches the last year of PubMed, filters for relevance and novelty, and produces a concise research summary plus optional slide-outline and email draft.
---

# Movement Science Lecture Update

Use this skill when a user wants to update an existing lecture deck with recent research while avoiding papers that are already redundant with the current lecture content.

## Inputs

Gather or confirm these inputs before starting:

- A lecture deck or lecture text source.
- The lecture identifier or topic to search around.
- Whether the user wants only a research summary, or also a slide outline and email draft.
- Optional delivery targets such as a Google Slides template, Drive folder, or email recipient.

If a deck is provided as Google Slides, extract slide text directly. If only text is available, work from that text.

## Core Goal

Find recent papers that are both:

- Relevant to the lecture objectives and key terms.
- Novel relative to what the current lecture already teaches.

Then produce a concise update package the instructor can use.

## Workflow

### 1. Extract the lecture baseline

Read the deck and identify:

- Lecture objectives.
- Key terms emphasized in the content.
- Concepts, mechanisms, definitions, and distinctions the lecture already covers.
- Activity or logistics slides that should be ignored.

Build a compact lecture baseline with fields like:

- `lecture_title`
- `known_points`
- `known_mechanisms`
- `known_definitions`
- `known_key_terms`
- `notes_for_novelty_check`

Do not invent facts beyond the deck.

### 2. Generate PubMed search queries

Using the lecture objectives and key terms, create multiple PubMed queries that are broad enough to find relevant recent work but specific enough to stay on-topic.

For each query, keep:

- `query`
- `rationale`

Prefer the last 12 months unless the user specifies a different range.

### 3. Search and normalize PubMed results

For each query:

- Search PubMed.
- Collect PMIDs.
- Deduplicate PMIDs across queries.
- Fetch article metadata and abstracts.
- Normalize each article into a consistent structure with:
  - `pmid`
  - `doi`
  - `title`
  - `abstract`
  - `journal`
  - `pubDate`
  - `authors`
  - `url`

Skip items with no usable abstract unless the user asks to include title-only records.

### 4. Score topical relevance

Evaluate each article against the lecture objectives and key terms.

Use this rule set:

- Score `0-100`.
- Mark `relevant=true` only when the paper clearly supports the lecture topic.
- Be strict about off-topic organisms, unrelated diseases, and weak conceptual overlap.

Return for each paper:

- `pmid`
- `score`
- `relevant`
- `reason`

Default threshold: keep only papers with relevance score `>= 75`.

### 5. Score novelty against the lecture baseline

For papers that passed relevance screening, score how much genuinely new information they add beyond the current lecture.

Use this rule set:

- Score `0-100`.
- Mark `novel=true` only when the abstract adds findings, mechanisms, distinctions, or evidence not already covered in the baseline.
- Low novelty means the paper mostly repeats concepts already taught.

Return for each paper:

- `pmid`
- `noveltyScore`
- `novel`
- `reason`
- `whatIsNew`

Default threshold: keep only papers with novelty score `>= 70`.

### 6. Produce the update package

Summarize only the papers that passed both screens.

Required outputs:

- A concise overall summary.
- Takeaways grouped by lecture objective when possible.
- A citation list with PMID, title, and URL.
- Notes on why the chosen papers matter.

Optional outputs when requested:

- A slide outline for a short lecture update deck.
- Replacement text or placeholders for a slide template.
- An email draft to the instructor summarizing the update.

## Output Contract

Unless the user asks for a different format, produce:

```json
{
  "summary": "string",
  "by_objective": [
    {
      "objective": "string",
      "takeaways": ["string"],
      "supporting_pmids": ["string"]
    }
  ],
  "citations": [
    {
      "pmid": "string",
      "title": "string",
      "url": "string"
    }
  ],
  "notes": ["string"]
}