{"id":"3518fce5-df88-4f35-a30b-aec1fb0cccad","shortId":"Ubpvg4","kind":"skill","title":"ma-scout","tagline":"Meta-analysis topic discovery and feasibility assessment. Professor-first (profile → gap) or Topic-first (question → gap → co-author). Pre-protocol phase from idea to ranked topic list.","description":"# MA Scout Skill\n\nYou are helping a medical researcher discover meta-analysis topics.\nTwo modes are available depending on the starting point.\n\nThis skill handles the **pre-protocol phase** — from idea to ranked topic list.\nFor actual MA execution (PROSPERO, screening, analysis), hand off to `/meta-analysis`.\n\n## Mode Selection\n\nDetermine the mode from user input:\n\n| Signal | Mode |\n|--------|------|\n| Professor name or profile URL provided | **A: Professor-first** |\n| Clinical question, keyword, trend, or \"find me a topic\" | **B: Topic-first** |\n| Both supplied (e.g., \"this topic with this professor\") | **A** (topic as filter) |\n\nIf ambiguous, ask the user whether to search by professor (supervisor-first) or by\ntopic (question-first).\n\n## Communication Rules\n\n- Communicate with the user in their preferred language (typically Korean).\n- Research questions, PICO/PIRD, and README content in English.\n- Medical terminology always in English.\n\n---\n\n## Inputs\n\n### Mode A: Professor-first\n- Professor name (native-language + English)\n- Profile URL (ScholarWorks, SKKU Faculty, Google Scholar, ORCID)\n- PubMed author link (preferably with cauthor_id for disambiguation)\n- Known specialty (e.g., \"thoracic imaging\", \"abdominal imaging\")\n- Affiliation history (e.g., \"Hospital A → Hospital B → retired\")\n- Minimum required: **name + at least one profile URL or PubMed link**\n\n### Mode B: Topic-first\n- Clinical question or keyword (e.g., \"AI for lung-nodule malignancy prediction\", \"dual-energy CT body composition\")\n- Radiology subspecialty scope (e.g., thoracic, abdominal, neuro)\n- MA type preference (DTA, prognostic, intervention — optional)\n- Desired role: solo first author / co-first / supervisor-matched\n- Minimum required: **clinical question or keyword**\n\n---\n\n## Workflow\n\n> **Mode A (Professor-first):** Phase 0 → 1 → 2 → 3 → 4 → 5\n> **Mode B (Topic-first):** T-Phase 0 → T-1 → T-2 → T-3 → T-4 → T-5\n> Phase 2 (MA Gap Analysis) and Phase 4 (README template) are shared between both modes.\n\n---\n\n# ═══════════════════════════════════════════\n# MODE A: PROFESSOR-FIRST WORKFLOW\n# ═══════════════════════════════════════════\n\n### Phase 0: Disambiguation & Context Confirmation\n\n**Goal:** Resolve author identity before any search, and confirm user's relationship context.\n\n**CRITICAL — Do this BEFORE any PubMed search:**\n\n1. **Resolve full English name first:**\n   - If cauthor_id is provided → fetch that specific PMID page to get full name + affiliation\n   - NEVER start with initials-only search (e.g., \"Ha HK\") — common Korean initials cause massive contamination\n   - First search must be `\"[Full Name]\"[Author]` (e.g., `\"Ha Hyun Kwon\"[Author]`)\n\n2. **Confirm affiliation chain with user:**\n   - Ask the user whether `{detected affiliation}` matches the professor's history,\n     and request the user's relationship to the professor so topic proposals can be\n     tuned accordingly.\n   - This prevents wrong-institution assumptions\n   - Skip only if user already provided explicit affiliation history\n\n3. **Profile URL fallback chain** (Scopus requires auth, so plan alternatives):\n   - 1st: PubMed full name search (always works)\n   - 2nd: Google Scholar profile (WebSearch `\"[Full Name]\" radiology scholar`)\n   - 3rd: ResearchGate profile (WebSearch `\"[Full Name]\" researchgate radiology`)\n   - 4th: ScholarWorks / SKKU / university faculty page (if URL provided)\n   - Last: Scopus/ScienceDirect (often fails due to auth — do NOT rely on it)\n\n---\n\n### Phase 1: Profile Exploration (E-utilities API)\n\n**Goal:** Identify the professor's 5-6 distinct research pillars using PubMed E-utilities API.\n\n**CRITICAL — Use E-utilities API, NOT WebFetch for PubMed:**\n- Scripts: `~/.claude/skills/search-lit/references/pubmed_eutils.sh` + `parse_pubmed.py`\n- Rate limit: 350ms between calls (100ms with NCBI_API_KEY)\n- These are faster, more reliable, and return structured data (JSON/XML)\n\n**Step 1 — Total publication count + PMID list:**\n```bash\nbash ~/.claude/skills/search-lit/references/pubmed_eutils.sh search \\\n  '\"[Full Name]\"[Author]' 200 \\\n  | python3 ~/.claude/skills/search-lit/references/parse_pubmed.py esearch\n```\n\n**Step 2 — Fetch metadata for MeSH-based clustering (parallel):**\n```bash\n# Get PMIDs from Step 1, then fetch summaries\nbash ~/.claude/skills/search-lit/references/pubmed_eutils.sh fetch_json \\\n  \"PMID1,PMID2,...\" \\\n  | python3 ~/.claude/skills/search-lit/references/parse_pubmed.py esummary\n```\n\n**Step 3 — Topic-specific counts (launch 4-5 searches in parallel via Bash):**\n```bash\n# Run these in parallel Bash calls\nbash ~/.claude/skills/search-lit/references/pubmed_eutils.sh search \\\n  '\"[Full Name]\"[Author] AND \"keyword1\"' 5\nbash ~/.claude/skills/search-lit/references/pubmed_eutils.sh search \\\n  '\"[Full Name]\"[Author] AND \"keyword2\"' 5\n# ... repeat for each suspected pillar keyword\n```\n\n**Step 4 — MeSH term extraction for automatic pillar clustering:**\n```bash\n# Fetch full XML for top-cited papers to extract MeSH headings\nbash ~/.claude/skills/search-lit/references/pubmed_eutils.sh fetch \\\n  \"PMID1,PMID2,...,PMID20\" \\\n  | python3 -c \"\nimport sys, xml.etree.ElementTree as ET\nfrom collections import Counter\nroot = ET.fromstring(sys.stdin.read())\nmesh_counts = Counter()\nfor article in root.findall('.//PubmedArticle'):\n    for mh in article.findall('.//MeshHeading/DescriptorName'):\n        mesh_counts[mh.text] += 1\nfor term, count in mesh_counts.most_common(30):\n    print(f'{count:3d}  {term}')\n\"\n```\n→ Top MeSH terms reveal natural research pillars (e.g., \"Colonography, Computed Tomographic\" = CTC pillar).\n\n**Step 5 — Google Scholar profile (parallel with PubMed calls):**\n- WebSearch: `\"[Full Name]\" radiology scholar google` for h-index, citation data\n\n**Output: Pillar Summary Table**\n\n| Pillar | 영역 | 대표 키워드 | MeSH terms | 추정 논문 수 |\n|--------|------|------------|-----------|-------------|\n| 1 | ... | ... | ... | ~N+ |\n\n---\n\n### Phase 2: MA Gap Analysis (Multi-Source)\n\n**Goal:** For each pillar, determine if a viable MA topic exists using PubMed + Consensus + Scholar Gateway + bioRxiv.\n\nFor each pillar (run in parallel using meta-analyst agents):\n\n#### 2a. PubMed E-utilities — Existing MAs + Primary studies\n\n```bash\n# Existing MAs (structured count)\nbash ~/.claude/skills/search-lit/references/pubmed_eutils.sh search \\\n  '[pillar keywords] AND (\"meta-analysis\"[pt] OR \"systematic review\"[pt])' 50\n\n# Primary studies with extractable outcomes\nbash ~/.claude/skills/search-lit/references/pubmed_eutils.sh search \\\n  '[pillar keywords] AND (\"sensitivity\" OR \"specificity\" OR \"accuracy\" OR \"prognosis\" OR \"outcome\")' 50\n```\n\n#### 2b. Consensus MCP — Semantic MA gap detection\n\nUse `mcp__claude_ai_Consensus__search` to find existing SRs/MAs that PubMed keyword search might miss:\n```\nquery: \"systematic review OR meta-analysis [pillar topic] [imaging modality]\"\n```\nConsensus returns citation-ranked results — check if any highly-cited MA already covers the proposed scope.\n**Limit:** max 3 Consensus calls per Phase 2 batch (rate limit).\n\n#### 2c. Scholar Gateway — Semantic similarity search\n\nUse `mcp__claude_ai_Scholar_Gateway__semanticSearch` for:\n- Finding MAs with different terminology (e.g., \"pooled analysis\" instead of \"meta-analysis\")\n- Detecting scope-overlapping MAs that use different keywords\n- Identifying methodological review papers that partially cover the topic\n\n#### 2d. bioRxiv/medRxiv — In-press competition detection\n\nUse `mcp__claude_ai_bioRxiv__search_preprints` to catch:\n- MAs posted as preprints but not yet indexed in PubMed\n- Ongoing SR/MA protocols shared as preprints\n- Very recent primary studies that could change feasibility\n\n```\nquery: \"[pillar keywords] meta-analysis OR systematic review\"\nserver: \"medrxiv\"  (for clinical topics)\n```\n\n#### 2e. Assessment matrix\n\n| Factor | Criteria |\n|--------|----------|\n| MA gap | 0 existing = best, 1-3 = check scope overlap, >5 = saturated |\n| Primary k | ≥8 for DTA, ≥6 for prognostic (minimum), ≥15 ideal |\n| Recency | Last MA >5 years old = update opportunity |\n| Competition | Check 2024-2026 for very recent MAs that block entry |\n\n#### 2f. PROSPERO competition check (MANDATORY)\n\n- Search PROSPERO via WebSearch: `site:crd.york.ac.uk/prospero [topic keywords]`\n- Also try WebFetch: `https://www.crd.york.ac.uk/prospero/#searchadvanced`\n- Look for registered-but-unpublished protocols that could block entry\n- If PROSPERO match found → flag as 🚫 competition risk in ranking\n\n#### 2g. Realistic k estimation\n\n- Raw PubMed hit count is NOT the real k — most studies lack 2x2 data or HR\n- Apply conservative discount: **k_realistic ≈ raw_count × 0.15–0.30** for DTA topics\n- Flag if k_realistic < 8 (DTA) or < 6 (prognostic) as ⚠️ feasibility risk\n- Report both raw and realistic estimates, e.g., `estimated k: ~130 (raw) → ~20–40 (extractable DTA data)`\n\n#### 2h. Niche subtopic discovery (if pillar appears saturated)\n\n- AI/radiomics angle on a classical topic\n- Specific modality comparison (e.g., CEUS vs MRI)\n- Treatment response (vs diagnosis which is often saturated)\n- Specific subpopulation or disease subtype\n- Use Consensus to check if the niche angle has already been covered\n\n---\n\n### Phase 3: Topic Ranking\n\n**Goal:** Rank all viable topics by composite score.\n\nScore each candidate on 5 criteria (★1-5):\n\n| Criteria | Weight | Description |\n|----------|--------|-------------|\n| **Professor fit** | Highest | Core area of the professor's career, publication count, distinctive contribution |\n| **MA gap** | High | No prior MA > ≥5 yr since last MA > recent MA exists |\n| **Feasibility (k)** | High | Number of includable studies and extractability of 2×2 or HR data |\n| **Clinical impact** | Medium | Whether the topic directly informs clinical decision-making |\n| **Execution ease** | Medium | Completable from literature alone; difficulty of managing heterogeneity |\n\n**Output: Ranked Topic Table**\n\n| Rank | Topic | Professor's Pillar | Prior MA | Estimated k (raw→realistic) | PROSPERO competition | Verdict |\n|------|-------|--------------------|----------|-----------------------------|----------------------|---------|\n| 1 | ... | ... | 0 | ~98 → 15–30 | None | ✅ Best fit |\n\n---\n\n### Phase 4: Folder & README Scaffolding\n\n**Goal:** Create project folders and README for each viable topic.\n\n1. **Folder location:** `{working_dir}/ma-scout/{initials}_{professor_name}/`\n2. **Naming convention:** `{NN}_{topic_slug}/` (within professor folder)\n   - Professor folder: `{initials}_{name}` (e.g., `KDK_Kim`, `LKS_Lee`)\n   - NN: sequential number within professor (01, 02, ...)\n   - topic_slug: English, underscore-separated\n   - Check existing folders with `ls` before creating\n\n3. **README.md template (PROSPERO-ready):**\n   Load the bilingual template block from\n   `${CLAUDE_SKILL_DIR}/references/project_readme_template.md` and copy it into\n   `{topic_folder}/README.md`. The reference covers both supervised (Mode A) and\n   solo-mode (Mode B, no supervisor) variants and contains the PICO/PIRD frame,\n   preliminary search, target journal table, and backward-planned timeline.\n\n---\n\n### Phase 5: Output Summary\n\n**Goal:** Persist findings for the user.\n\n1. Save the ranked topic table and README files to the working directory.\n2. Summarize: total topics scanned, viable topics found, recommended next steps.\n3. Suggest the user save results to their project management system (e.g., `/manage-project`).\n\n---\n\n## Niche Topic Discovery Heuristics\n\nWhen all major pillars are saturated (>5 prior MAs), try these angles:\n\n1. **\"First MA\" rule:** Professor's most unique/niche subtopic where MA = 0\n2. **AI/radiomics overlay:** Classical imaging topic + AI approach = new MA angle\n3. **Treatment response:** Diagnosis MAs saturated → treatment monitoring MA often open\n4. **Modality comparison:** Head-to-head (e.g., CEUS vs MRI) often underserved\n5. **Guideline gap:** Professor authored guidelines → MA supporting/updating those guidelines\n6. **Geographic/population niche:** Regional population-specific MA (e.g., parasitic diseases, TB)\n7. **Temporal update:** Last MA >5 years old + significant new primary studies since\n\n---\n\n## Quality Gates\n\nBefore finalizing a topic as viable:\n\n- [ ] **Author identity confirmed** — full name resolved via E-utilities efetch, no initials-only contamination\n- [ ] **Affiliation confirmed** with user (or from reliable source)\n- [ ] Confirmed MA = 0 or last MA >5 years (via PubMed E-utilities, not assumption)\n- [ ] **Cross-validated via Consensus/Scholar Gateway** — no hidden MAs with different terminology\n- [ ] **bioRxiv/medRxiv checked** — no preprint MA in progress\n- [ ] Confirmed k_realistic ≥ 8 (DTA) or ≥ 6 (prognostic) after discount\n- [ ] **PROSPERO searched** — no registered competing protocol found\n- [ ] No 2024-2026 competing MA in press (check PubMed + preprints)\n- [ ] Professor's publication record demonstrates clear authority in this area\n- [ ] Research question is specific enough for PROSPERO registration\n- [ ] **README contains:** complete PICO/PIRD, PubMed search strategy, Embase draft, target journal with IF, timeline\n\n---\n\n## Handoff\n\nAfter MA Scout completes:\n- To **`/meta-analysis`**: When a topic is approved and ready for PROSPERO protocol (README has PICO + search strategy ready)\n- To **`manage-project`**: When project folder needs full scaffolding\n- To **`search-lit`**: When deeper preliminary search is needed before committing\n- To **`/analyze-stats`**: When feasibility requires power/sample-size calculation for the estimated k\n\n---\n\n## Parallel Execution Strategy\n\nFor efficiency, launch multiple agents and API calls in parallel:\n\n**Phase 0 (Identity):**\n1. E-utilities esearch: `\"[Full Name]\"[Author]` → total count + PMIDs (FIRST)\n2. E-utilities efetch: top 20 PMIDs → MeSH terms → automatic pillar clustering\n\n**Phase 1 (Profile — all parallel):**\n3. Bash × 4-5: E-utilities esearch with topic-specific filters (parallel Bash calls)\n4. WebSearch: Google Scholar profile\n5. WebFetch: any provided profile URLs (skip Scopus)\n\n**Phase 2 (MA Gap — multi-source parallel):**\n6. Up to 4 meta-analyst agents in parallel, each covering 1-2 pillars\n7. Each agent runs ALL of:\n   - E-utilities esearch: existing MA count + primary study count\n   - Consensus MCP: semantic MA search (max 3 calls total across all agents)\n   - Scholar Gateway: scope-overlap check\n   - bioRxiv/medRxiv: preprint MA detection\n   - PROSPERO: competition check (WebSearch)\n8. Each agent reports: raw k, realistic k (15-30% discount), all sources checked\n\n**Phase 3 (Ranking):** Sequential, uses Phase 2 outputs.\n\n**Phase 4 (Scaffolding):** Sequential, creates folders + PROSPERO-ready READMEs.\n\nTotal (Mode A): 5-8 parallel agents per professor, ~8-12 minutes per professor.\n\n### Mode B Parallel Strategy\n\n**T-Phase 0:** Sequential (user interaction for scope clarification).\n\n**T-Phase 1 (Landscape — all angles in parallel):**\n1. Per angle: Bash (PubMed MA count) + Bash (primary k) + Consensus + bioRxiv + PROSPERO\n2. 3-5 angles × 5 sources = 15-25 parallel calls\n\n**T-Phase 2 (Deep-dive):** Same as Mode A Phase 2, only for viable angles (typically 1-2).\n\n**T-Phase 4 (Co-author — if needed):**\n3. Bash: PubMed author frequency search\n4. WebSearch: Google Scholar profiles for top candidates\n\nTotal (Mode B): ~5-8 minutes per topic scan (faster than Mode A — no profile exploration).\n\n### Known Pitfalls (from 3 professor analyses)\n- Common Korean/Asian initials (e.g., \"Lee KS\", \"Kim DK\") return 300+ papers with massive contamination. Always use full name first.\n- Scopus/ScienceDirect → 403 or redirect to login. Never rely on Scopus as primary data source.\n- Raw PubMed counts overestimate by 3-7x. ~130 hits often means 20-40 with extractable DTA data.\n- Professor may have moved institutions. Don't assume affiliation without verification.\n- **Consensus rate limit:** Max 3 batch calls. If rate-limited, wait 30s and retry once.\n- **E-utilities rate limit:** 350ms between calls (100ms with NCBI_API_KEY). Scripts handle this automatically.\n- **bioRxiv MCP:** Use `server: \"medrxiv\"` for clinical topics, `server: \"biorxiv\"` for preclinical.\n\n---\n\n# ═══════════════════════════════════════════\n# MODE B: TOPIC-FIRST WORKFLOW\n# ═══════════════════════════════════════════\n\n### T-Phase 0: Topic Clarification & Scope\n\n**Goal:** Refine the user's clinical question into a searchable, PROSPERO-registrable scope.\n\n1. **Parse the input** — extract:\n   - Disease/condition (e.g., \"lung nodule\", \"hepatocellular carcinoma\")\n   - Imaging modality or intervention (e.g., \"dual-energy CT\", \"AI CAD\")\n   - Outcome type: DTA (Se/Sp), prognostic (HR/OR), intervention (RR/MD), dosimetry\n   - Population specifics (e.g., \"screening setting\", \"cirrhotic patients\")\n\n2. **Expand to neighboring angles** — propose 3-5 variations:\n   ```\n   user input: \"AI for lung nodule malignancy prediction\"\n   → variant 1: AI vs radiologist for lung nodule malignancy prediction (DTA)\n   → variant 2: Radiomics for lung nodule malignancy (DTA)\n   → variant 3: Deep learning for incidental pulmonary nodule management (prognostic)\n   → variant 4: AI-assisted Lung-RADS upgrade accuracy (DTA)\n   → variant 5: Low-dose CT AI for lung cancer screening (DTA)\n   ```\n\n3. **User selects 1-3 angles** to investigate further.\n\n---\n\n### T-Phase 1: Landscape Scan (Multi-Source)\n\n**Goal:** For each selected angle, rapidly assess the MA landscape.\n\n**Run all angles in parallel. For each angle:**\n\n#### 1a. PubMed — Existing MA count\n```bash\nbash ~/.claude/skills/search-lit/references/pubmed_eutils.sh search \\\n  '[topic keywords] AND (\"meta-analysis\"[pt] OR \"systematic review\"[pt])' 50\n```\n\n#### 1b. PubMed — Primary study pool\n```bash\nbash ~/.claude/skills/search-lit/references/pubmed_eutils.sh search \\\n  '[topic keywords] AND (\"sensitivity\" OR \"specificity\" OR \"hazard\" OR \"outcome\")' 100\n```\n\n#### 1c. Consensus MCP — Semantic MA discovery\n```\nquery: \"systematic review [topic] [modality]\"\n```\nCheck for MAs using different terminology.\n\n#### 1d. bioRxiv/medRxiv — Preprint competition\n```\nquery: \"[topic] meta-analysis\"\nserver: \"medrxiv\"\n```\n\n#### 1e. PROSPERO — Registered protocols\nWebSearch: `site:crd.york.ac.uk/prospero [topic keywords]`\n\n**Output: Landscape Summary Table**\n\n| 변형 | 기존 MA | Primary k (raw) | k (realistic) | PROSPERO | Preprint MA | 판정 |\n|------|---------|----------------|--------------|----------|------------|------|\n| 1 | 3편 | 120 | 18-36 | 1건 | 0 | ⚠️ 경쟁 |\n| 2 | 0편 | 85 | 13-25 | 0 | 0 | ✅ 최적 |\n\n---\n\n### T-Phase 2: Feasibility Deep-Dive\n\n**Goal:** For viable angles (MA ≤ 2, no PROSPERO conflict), run full gap analysis.\n\nThis phase uses the **same Phase 2 (MA Gap Analysis)** as Mode A — steps 2a through 2h.\nThe only difference: no \"Professor fit\" to evaluate, so focus on:\n- **Gap certainty** — are existing MAs truly non-overlapping with proposed scope?\n- **k quality** — are primary studies heterogeneous enough to warrant MA, or too uniform?\n- **User's domain fit** — does this align with user's radiology AI / imaging expertise?\n\n---\n\n### T-Phase 3: Topic Ranking (Topic-first weights)\n\n**Goal:** Rank viable topics with weights adjusted for topic-first approach.\n\n| Criteria | Weight | Description |\n|----------|--------|-------------|\n| **MA gap** | 최고 | 기존 MA 없음 > update 기회 > 포화 |\n| **Feasibility (k)** | 최고 | k_realistic ≥ 8 (DTA) or ≥ 6 (prognostic) |\n| **User domain fit** | 높음 | 사용자의 전문 분야와 맞는가 |\n| **Clinical impact** | 중간 | 가이드라인 변경 가능성, 임상 의사결정 직결 |\n| **Co-author availability** | 중간 | 해당 분야 전문가 접근 가능성 (기존 관계 or 접근 용이) |\n| **Execution ease** | 중간 | 단독 진행 가능 vs 전문가 해석 필수 |\n\n**Output: Ranked Topic Table**\n\n| 순위 | 주제 | 기존 MA | 추정 k | PROSPERO | Co-author 필요 | 종합 |\n|------|------|---------|--------|----------|--------------|------|\n| 1 | ... | 0편 | 25 | 없음 | 선택적 | ✅ 최적 |\n\n---\n\n### T-Phase 4: Co-Author Matching (Optional)\n\n**Goal:** If the user wants a senior co-author, find candidates.\n\n**Strategy 1 — Existing network (memory-based):**\n- Check memory files for professors with overlapping expertise\n- Cross-reference existing professor folders in the working directory\n- Best match = professor whose pillar naturally covers this topic\n\n**Strategy 2 — PubMed reverse search:**\n```bash\n# Find prolific authors in this specific topic\nbash ~/.claude/skills/search-lit/references/pubmed_eutils.sh search \\\n  '[topic keywords] AND (\"{user_country}\"[Affiliation])' 100\n```\nThen:\n- E-utilities efetch → extract author frequency from results\n- Top 5 most-published authors in this niche = potential co-authors\n- Cross-check Google Scholar for h-index and recent activity\n\n**Strategy 3 — Self-led (no senior co-author):**\n- Viable when: user has 2+ published MAs, topic is methodologically straightforward\n- Still need 2nd reviewer (junior colleague or peer) — flag this in README\n- Corresponding author = user\n\n**Output:** Co-author recommendation table or a \"solo-viable\" judgment.\n\n---\n\n### T-Phase 5: Folder & README Scaffolding (Topic-first)\n\n**Goal:** Create project folder and PROSPERO-ready README.\n\n1. **Folder location:** `{working_dir}/ma-scout/TOPIC/`\n   - Topic-first projects use `TOPIC/` prefix (not professor initials)\n   - Naming: `{NN}_{Topic_Abbreviation}/` (e.g., `01_AI_Lung_Nodule_DTA/`)\n   - If co-author matched later, can be moved under professor folder\n\n2. **README.md template:** Same PROSPERO-ready template as Mode A Phase 4 (see\n   `references/project_readme_template.md`), with these changes:\n   - `Supervisor:` → `Lead: {user_name}` or `Lead: {user_name} + {co-author}`\n   - Drop the supervisor-area row; use `Domain: {subspecialty}` instead.\n   - Rename `Professor's Authority` → `Team Expertise` (user's credentials + co-author if any)\n   - Timeline: drop the supervisor-proposal step → start directly at PROSPERO registration.\n\n   **Timeline template (self-led):**\n   | Step | Expected timing | Precondition |\n   |------|-----------------|--------------|\n   | PROSPERO registration | {YYYY-MM} | topic confirmed |\n   | Search complete | +1 week | PROSPERO registration |\n   | Screening complete | +2 weeks | 2nd reviewer secured |\n   | Data extraction | +3 weeks | screening consensus |\n   | Analysis + draft | +5 weeks | data lock |\n   | Co-author review | +7 weeks | draft complete |\n   | Submission | +8 weeks | final approval |\n\n3. **Summary:** Same as Mode A Phase 5 — save ranked results and recommend next steps.\n\n---\n\n### Topic Discovery Heuristics (Mode B specific)\n\nWhen the user asks for topic suggestions without a specific idea:\n\n1. **Trend scan** — Search recent high-IF radiology journals for \"gap in the literature\" + \"meta-analysis needed\":\n   ```bash\n   bash ~/.claude/skills/search-lit/references/pubmed_eutils.sh search \\\n     '\"no meta-analysis\" AND \"radiology\"[Journal] AND 2024:2026[dp]' 30\n   ```\n\n2. **Guideline update gaps** — New guidelines (ACR, ESR, RSNA) often cite lack of MA evidence:\n   - Consensus search: `\"practice guideline\" AND \"insufficient evidence\" AND [radiology subspecialty]`\n\n3. **AI + classical imaging** — Overlay AI/DL/radiomics on well-studied classical topics:\n   - Many classical DTA topics have 10+ MAs, but AI angle has 0-1\n\n4. **Korean/Asian population** — Population-specific MA for diseases with geographic variation:\n   - TB, NTM, parasitic diseases, gastric cancer, liver fluke, HBV-related HCC\n\n5. **Technology adoption** — New modalities with growing evidence but no synthesis:\n   - Photon-counting CT, abbreviated MRI, contrast-enhanced mammography, AI CAD\n\n6. **Cross-subspecialty** — Topics spanning two subspecialties often fall through MA cracks:\n   - Cardiac + thoracic (coronary CT + lung screening), neuro + MSK (spine imaging)\n\n---\n\n### Quality Gates (Mode B specific)\n\nBefore finalizing a topic-first MA as viable:\n\n- [ ] Clinical question refined to PICO/PIRD (not just a keyword)\n- [ ] MA gap confirmed via PubMed + Consensus + Scholar Gateway + bioRxiv (all 4 sources)\n- [ ] k_realistic ≥ 8 (DTA) or ≥ 6 (prognostic) after 15-30% discount\n- [ ] PROSPERO searched — no competing registered protocol\n- [ ] No 2024-2026 competing MA in press or preprint\n- [ ] User's domain expertise sufficient for clinical interpretation (or co-author identified)\n- [ ] 2nd reviewer identified or plan to recruit\n- [ ] README contains: complete PICO/PIRD, PubMed + Embase search strategy, target journal with IF, timeline\n- [ ] If self-led: user has ≥ 2 published MAs (otherwise, recommend co-author)\n\n---\n\n## Phase 6: Pre-Proposal Pipeline (Post-Scout)\n\nAfter MA Scout identifies viable topics, run the **pre-proposal pipeline** to prepare\na \"ready-to-propose\" package before contacting the professor.\n\n### Pipeline Steps\n\n1. **Search Execution** — E-utilities with broadened synonyms (retmax=200)\n   - Primary search: `[topic] AND [outcome keywords]`\n   - Existing MA search: `[topic] AND (\"meta-analysis\"[pt] OR \"systematic review\"[pt])`\n\n2. **Metadata Collection** — `fetch_json` → `esummary` (batch 40-50 PMIDs)\n\n3. **Title-Based Triage** — Classify as INCLUDE / MAYBE / EXCLUDE\n   - CRITICAL: Check for existing MAs within results (initial scout may miss them)\n   - Separate bronchoscopic vs percutaneous (30% contamination in CBCT topics)\n   - Flag professor's own papers (authority evidence)\n   - Flag retracted papers\n\n4. **PRISMA Flow Draft** — Identification → Screening → Eligibility → Included (estimated)\n\n5. **Gap Re-assessment** — Update MA count, re-position if needed:\n   - MA=0 → \"first MA\" | MA=1 (>5yr) → \"update MA\" | MA≥3 (recent) → skip/niche\n\n6. **Output Files**:\n   - `candidates.md` — full triage table + PRISMA flow + gap finding\n   - `README.md` — updated Preliminary Search section with actual numbers\n\n### Parallel Execution\n- Launch up to 4 agents per wave (each topic independent)\n- Each agent: search → fetch → triage → write files → ~5-10 min\n- 21 topics completed in ~1 hour with 16 parallel agents\n\n### Professor Contact Package\nThe pre-proposal gives the professor:\n- Candidate count + gap evidence (e.g., \"MA = 0, 35 studies to include\")\n- Clear role description (e.g., \"independent screening review + discussion only\")\n- Urgency of PROSPERO pre-registration to secure the topic\n\n## Anti-Hallucination\n\n- **Never fabricate publication counts, h-index, or pillar classifications.** All numbers must come from PubMed E-utilities API output.\n- **Never fabricate existing MA counts.** Always verify via PubMed search + PROSPERO check before claiming \"MA = 0\".\n- **Never invent professor expertise or affiliation.** Confirm with user before proceeding.\n- **k_realistic must use the 15-30% discount.** Raw PubMed counts overestimate by 3-7x. Always report both raw and realistic estimates.\n- If PubMed returns 0 or Consensus/Scholar Gateway is unavailable, state the limitation rather than guessing.","tags":["scout","medsci","skills","aperivue","agent-skills","biostatistics","claude-code","claude-skills","clinical-research","diagnostic-accuracy","irb-protocol","literature-review"],"capabilities":["skill","source-aperivue","skill-ma-scout","topic-agent-skills","topic-biostatistics","topic-claude-code","topic-claude-skills","topic-clinical-research","topic-diagnostic-accuracy","topic-irb-protocol","topic-literature-review","topic-manuscript","topic-medical-ai","topic-medical-research","topic-meta-analysis"],"categories":["medsci-skills"],"synonyms":[],"warnings":[],"endpointUrl":"https://skills.sh/Aperivue/medsci-skills/ma-scout","protocol":"skill","transport":"skills-sh","auth":{"type":"none","details":{"cli":"npx skills add Aperivue/medsci-skills","source_repo":"https://github.com/Aperivue/medsci-skills","install_from":"skills.sh"}},"qualityScore":"0.499","qualityRationale":"deterministic score 0.50 from registry signals: · indexed on github topic:agent-skills · 98 github stars · SKILL.md body (26,675 chars)","verified":false,"liveness":"unknown","lastLivenessCheck":null,"agentReviews":{"count":0,"score_avg":null,"cost_usd_avg":null,"success_rate":null,"latency_p50_ms":null,"narrative_summary":null,"summary_updated_at":null},"enrichmentModel":"deterministic:skill-github:v1","enrichmentVersion":1,"enrichedAt":"2026-05-18T18:56:30.338Z","embedding":null,"createdAt":"2026-05-13T12:57:45.393Z","updatedAt":"2026-05-18T18:56:30.338Z","lastSeenAt":"2026-05-18T18:56:30.338Z","tsv":"'+1':2964 '+2':2970 '+3':2977 '+5':2983 '+7':2991 '+8':2996 '-1':305,3116 '-10':3485 '-12':1963 '-2':307,1877,2032 '-2026':1068,1685,3241 '-25':2010,2463 '-3':309,1040,2323 '-30':1930,3231,3594 '-36':2455 '-4':311 '-40':2124 '-5':313,624,1243,1830,2005,2257 '-50':3368 '-6':527 '-7':2117,3602 '-8':1957,2060 '/.claude/skills/search-lit/references/parse_pubmed.py':586,614 '/.claude/skills/search-lit/references/pubmed_eutils.sh':548,579,608,638,647,684,832,852,2362,2383,2732,3053 '/analyze-stats':1771 '/ma-scout':1359 '/ma-scout/topic':2848 '/manage-project':1501 '/meshheading/descriptorname':715 '/meta-analysis':83,1731 '/prospero':1088,2432 '/prospero/#searchadvanced':1096 '/pubmedarticle':710 '/readme.md':1423 '/references/project_readme_template.md':1416 '0':289,303,336,1036,1332,1529,1634,1795,1974,2194,2457,2464,2465,3115,3434,3513,3576,3614 '0.15':1145 '0.30':1146 '01':1386,2864 '02':1387 '0편':2460,2658 '1':290,360,514,571,603,719,779,1039,1242,1331,1354,1465,1518,1797,1823,1876,1984,1990,2031,2212,2268,2322,2331,2451,2657,2685,2843,3032,3330,3438,3491 '10':3109 '100':2395,2740 '100ms':555,2164 '120':2453 '13':2462 '130':1171,2119 '15':1055,1334,1929,2009,3230,3593 '16':3494 '18':2454 '1a':2355 '1b':2376 '1c':2396 '1d':2413 '1e':2424 '1st':468 '1건':2456 '2':291,315,409,589,782,926,1285,1286,1363,1478,1530,1809,1857,1941,2003,2016,2025,2250,2279,2459,2470,2480,2494,2719,2790,2881,3067,3287,3360 '20':1173,1815,2123 '200':584,3340 '2024':1067,1684,3063,3240 '2026':3064 '21':3487 '25':2659 '2a':817,2502 '2b':867 '2c':930 '2d':975 '2e':1029 '2f':1076 '2g':1118 '2h':1178,2504 '2nd':475,2799,2972,3261 '2x2':1134 '3':292,457,617,921,1225,1401,1489,1541,1827,1901,1936,2004,2042,2075,2116,2144,2256,2287,2319,2558,2777,3000,3092,3370,3443,3601 '30':726,1335,3066,3396 '300':2087 '30s':2152 '35':3514 '350ms':552,2161 '3d':730 '3rd':484 '3편':2452 '4':293,321,623,662,1340,1552,1829,1843,1867,1944,2036,2048,2297,2666,2893,3117,3220,3411,3470 '40':1174,3367 '403':2098 '4th':492 '5':294,526,645,654,746,1044,1060,1240,1267,1456,1512,1565,1592,1638,1848,1956,2007,2059,2308,2752,2827,3007,3141,3420,3484 '50':845,866,2375 '5yr':3439 '6':1051,1157,1575,1672,1864,2597,3164,3227,3296,3446 '7':1587,1879 '8':1048,1154,1669,1921,1962,2594,3224 '85':2461 '98':1333 'abbrevi':2862,3156 'abdomin':207,256 'accord':441 'accuraci':861,2305 'acr':3073 'across':1904 'activ':2775 'actual':74,3463 'adjust':2571 'adopt':3143 'affili':209,380,411,420,455,1624,2137,2739,3582 'agent':816,1788,1871,1881,1906,1923,1959,3471,3478,3496 'ai':238,877,939,985,1536,2232,2261,2269,2299,2313,2552,2865,3093,3112,3162 'ai-assist':2298 'ai/dl/radiomics':3097 'ai/radiomics':1186,1531 'align':2547 'alon':1308 'alreadi':452,914,1221 'also':1091 'altern':467 'alway':170,473,2092,3566,3604 'ambigu':130 'analys':2077 'analysi':6,48,79,318,785,839,896,951,956,1020,2369,2421,2487,2497,2981,3049,3058,3354 'analyst':815,1870 'angl':1187,1219,1517,1540,1987,1992,2006,2029,2254,2324,2341,2349,2354,2478,3113 'anti':3538 'anti-hallucin':3537 'api':520,536,542,558,1790,2167,3559 'appear':1184 'appli':1138 'approach':1537,2576 'approv':1736,2999 'area':1251,1702,2914 'articl':707 'article.findall':714 'ask':131,415,3024 'assess':11,1030,2343,3424 'assist':2300 'assum':2136 'assumpt':447,1646 'auth':464,507 'author':25,194,269,342,403,408,583,642,651,1569,1608,1699,1804,2039,2045,2618,2654,2669,2681,2726,2747,2756,2763,2785,2810,2815,2872,2909,2923,2931,2989,3259,3294,3406 'automat':667,1819,2172 'avail':53,2619 'b':113,215,229,296,1436,1968,2058,2186,3019,3190 'backward':1452 'backward-plan':1451 'base':595,2690,3373 'bash':577,578,598,607,629,630,635,637,646,670,683,826,831,851,1828,1841,1993,1997,2043,2360,2361,2381,2382,2723,2731,3051,3052 'batch':927,2145,3366 'best':1038,1337,2709 'bilingu':1409 'biorxiv':805,986,2001,2173,2182,3218 'biorxiv/medrxiv':976,1659,1913,2414 'block':1074,1106,1411 'bodi':249 'broaden':3337 'bronchoscop':3393 'c':690 'cad':2233,3163 'calcul':1776 'call':554,636,753,923,1791,1842,1902,2012,2146,2163 'cancer':2316,3134 'candid':1238,2055,2683,3507 'candidates.md':3449 'carcinoma':2222 'cardiac':3177 'career':1256 'catch':990 'caus':394 'cauthor':198,367 'cbct':3399 'certainti':2517 'ceus':1196,1560 'chain':412,461 'chang':1013,2898 'check':907,1041,1066,1079,1215,1394,1660,1690,1912,1919,1934,2407,2691,2766,3381,3572 'cirrhot':2248 'citat':764,904 'citation-rank':903 'cite':677,912,3077 'claim':3574 'clarif':1980,2196 'classic':1190,1533,3094,3102,3105 'classif':3549 'classifi':3375 'claud':876,938,984,1413 'clear':1698,3518 'clinic':104,233,278,1027,1290,1298,2179,2203,2607,3201,3254 'cluster':596,669,1821 'co':24,271,2038,2617,2653,2668,2680,2762,2784,2814,2871,2908,2930,2988,3258,3293 'co-author':23,2037,2616,2652,2667,2679,2761,2783,2813,2870,2907,2929,2987,3257,3292 'co-first':270 'colleagu':2802 'collect':697,3362 'colonographi':740 'come':3553 'commit':1769 'common':391,725,2078 'communic':148,150 'comparison':1194,1554 'compet':1680,1686,3236,3242 'competit':980,1065,1078,1114,1329,1918,2416 'complet':1305,1713,1729,2963,2969,2994,3270,3489 'composit':250,1234 'comput':741 'confirm':339,348,410,1610,1625,1632,1666,2961,3212,3583 'conflict':2483 'consensus':802,868,878,901,922,1213,1895,2000,2140,2397,2980,3082,3215 'consensus/scholar':1651,3616 'conserv':1139 'contact':3325,3498 'contain':1441,1712,3269 'contamin':396,1623,2091,3397 'content':165 'context':338,352 'contrast':3159 'contrast-enhanc':3158 'contribut':1260 'convent':1365 'copi':1418 'core':1250 'coronari':3179 'correspond':2809 'could':1012,1105 'count':574,621,704,717,722,729,830,1125,1144,1258,1806,1891,1894,1996,2113,2359,3154,3427,3508,3543,3565,3598 'counter':699,705 'countri':2738 'cover':915,972,1223,1426,1875,2715 'crack':3176 'crd.york.ac.uk':1087,2431 'crd.york.ac.uk/prospero':1086,2430 'creat':1345,1400,1947,2835 'credenti':2928 'criteria':1033,1241,1244,2577 'critic':353,537,3380 'cross':1648,2700,2765,3166 'cross-check':2764 'cross-refer':2699 'cross-subspecialti':3165 'cross-valid':1647 'ct':248,2231,2312,3155,3180 'ctc':743 'data':568,765,1135,1177,1289,2109,2128,2975,2985 'decis':1300 'decision-mak':1299 'deep':2018,2288,2473 'deep-div':2017,2472 'deeper':1763 'demonstr':1697 'depend':54 'descript':1246,2579,3520 'desir':265 'detect':419,873,957,981,1916 'determin':86,793 'diagnosi':1202,1544 'differ':947,964,1657,2411,2507 'difficulti':1309 'dir':1358,1415,2847 'direct':1296,2942 'directori':1477,2708 'disambigu':201,337 'discount':1140,1675,1931,3232,3595 'discov':45 'discoveri':8,1181,1504,2401,3016 'discuss':3525 'diseas':1210,1585,3125,3132 'disease/condition':2217 'distinct':528,1259 'dive':2019,2474 'dk':2085 'domain':2543,2600,2917,3250 'dose':2311 'dosimetri':2242 'dp':3065 'draft':1719,2982,2993,3414 'drop':2910,2935 'dta':261,1050,1148,1155,1176,1670,2127,2236,2277,2285,2306,2318,2595,2868,3106,3225 'dual':246,2229 'dual-energi':245,2228 'due':505 'e':518,534,540,820,1616,1643,1799,1811,1832,1886,2157,2743,3334,3557 'e-util':517,533,539,819,1615,1642,1798,1810,1831,1885,2156,2742,3333,3556 'e.g':119,204,211,237,254,388,404,739,949,1168,1195,1376,1500,1559,1583,2081,2218,2227,2245,2863,3511,3521 'eas':1303,2632 'efetch':1618,1813,2745 'effici':1785 'elig':3417 'embas':1718,3273 'energi':247,2230 'english':167,172,184,363,1390 'enhanc':3160 'enough':1707,2534 'entri':1075,1107 'esearch':587,1801,1834,1888 'esr':3074 'estim':1121,1167,1169,1324,1779,3419,3610 'esummari':615,3365 'et':695 'et.fromstring':701 'evalu':2512 'evid':3081,3088,3148,3407,3510 'exclud':3379 'execut':76,1302,1782,2631,3332,3466 'exist':799,822,827,882,1037,1274,1395,1889,2357,2519,2686,2702,3347,3383,3563 'expand':2251 'expect':2952 'expertis':2554,2698,2925,3251,3580 'explicit':454 'explor':516,2071 'extract':665,680,849,1175,1283,2126,2216,2746,2976 'f':728 'fabric':3541,3562 'factor':1032 'faculti':189,496 'fail':504 'fall':3173 'fallback':460 'faster':562,2065 'feasibl':10,1014,1160,1275,1773,2471,2589 'fetch':371,590,605,609,671,685,3363,3480 'file':1473,2693,3448,3483 'filter':128,1839 'final':1603,2998,3193 'find':109,881,944,1461,2682,2724,3456 'first':14,20,103,116,141,147,178,232,268,272,287,299,333,365,397,1519,1808,2096,2189,2563,2575,2833,2851,3197,3435 'fit':1248,1338,2510,2544,2601 'flag':1112,1150,2805,3401,3408 'flow':3413,3454 'fluke':3136 'focus':2514 'folder':1341,1347,1355,1371,1373,1396,1422,1754,1948,2704,2828,2837,2844,2880 'found':1111,1485,1682 'frame':1444 'frequenc':2046,2748 'full':362,378,401,470,480,488,581,640,649,672,755,1611,1756,1802,2094,2485,3450 'gap':16,22,317,784,872,1035,1262,1567,1859,2486,2496,2516,2581,3043,3070,3211,3421,3455,3509 'gastric':3133 'gate':1601,3188 'gateway':804,932,941,1652,1908,3217,3617 'geograph':3127 'geographic/population':1576 'get':377,599 'give':3504 'goal':340,521,789,1228,1344,1459,2198,2337,2475,2565,2672,2834 'googl':190,476,747,759,1845,2050,2767 'grow':3147 'guess':3625 'guidelin':1566,1570,1574,3068,3072,3085 'h':762,2771,3545 'h-index':761,2770,3544 'ha':389,405 'hallucin':3539 'hand':80 'handl':61,2170 'handoff':1725 'hazard':2392 'hbv':3138 'hbv-relat':3137 'hcc':3140 'head':682,1556,1558 'head-to-head':1555 'help':41 'hepatocellular':2221 'heterogen':1312,2533 'heurist':1505,3017 'hidden':1654 'high':911,1263,1277,3038 'high-if':3037 'highest':1249 'highly-cit':910 'histori':210,425,456 'hit':1124,2120 'hk':390 'hospit':212,214 'hour':3492 'hr':1137,1288 'hr/or':2239 'hyun':406 'id':199,368 'idea':31,68,3031 'ideal':1056 'ident':343,1609,1796 'identif':3415 'identifi':522,966,3260,3263,3307 'imag':206,208,899,1534,2223,2553,3095,3186 'impact':1291,2608 'import':691,698 'in-press':977 'incident':2291 'includ':1280,3377,3418,3517 'independ':3476,3522 'index':763,998,2772,3546 'inform':1297 'initi':385,393,1360,1374,1621,2080,2858,3387 'initials-on':384,1620 'input':91,173,2215,2260 'instead':952,2919 'institut':446,2133 'insuffici':3087 'interact':1977 'interpret':3255 'intervent':263,2226,2240 'invent':3578 'investig':2326 'journal':1448,1721,3041,3061,3277 'json':610,3364 'json/xml':569 'judgment':2823 'junior':2801 'k':1047,1120,1130,1141,1152,1170,1276,1325,1667,1780,1926,1928,1999,2443,2445,2528,2590,2592,2650,3222,3588 'kdk':1377 'key':559,2168 'keyword':106,236,281,660,835,855,886,965,1017,1090,2365,2386,2434,2735,3209,3346 'keyword1':644 'keyword2':653 'kim':1378,2084 'known':202,2072 'korean':159,392 'korean/asian':2079,3118 'ks':2083 'kwon':407 'lack':1133,3078 'landscap':1985,2332,2346,2436 'languag':157,183 'last':501,1058,1270,1590,1636 'later':2874 'launch':622,1786,3467 'lead':2900,2904 'learn':2289 'least':221 'led':2780,2950,3284 'lee':1380,2082 'limit':551,919,929,2142,2150,2160,3622 'link':195,227 'list':35,72,576 'lit':1761 'literatur':1307,3046 'liver':3135 'lks':1379 'load':1407 'locat':1356,2845 'lock':2986 'login':2102 'look':1097 'low':2310 'low-dos':2309 'ls':1398 'lung':241,2219,2263,2273,2282,2302,2315,2866,3181 'lung-nodul':240 'lung-rad':2301 'ma':2,36,75,258,316,783,797,871,913,1034,1059,1261,1266,1271,1273,1323,1520,1528,1539,1549,1571,1582,1591,1633,1637,1663,1687,1727,1858,1890,1898,1915,1995,2345,2358,2400,2441,2449,2479,2495,2537,2580,2584,2648,3080,3123,3175,3198,3210,3243,3305,3348,3426,3433,3436,3437,3441,3442,3512,3564,3575 'ma-scout':1 'major':1508 'make':1301 'malign':243,2265,2275,2284 'mammographi':3161 'manag':1311,1498,1750,2294 'manage-project':1749 'mandatori':1080 'mani':3104 'mas':823,828,945,961,991,1072,1514,1545,1655,2409,2520,2792,3110,3289,3384 'massiv':395,2090 'match':275,421,1110,2670,2710,2873 'matrix':1031 'max':920,1900,2143 'may':2130,3389 'mayb':3378 'mcp':869,875,937,983,1896,2174,2398 'mean':2122 'medic':43,168 'medium':1292,1304 'medrxiv':1025,2177,2423 'memori':2689,2692 'memory-bas':2688 'mesh':594,663,681,703,716,733,774,1817 'mesh-bas':593 'mesh_counts.most':724 'meta':5,47,814,838,895,955,1019,1869,2368,2420,3048,3057,3353 'meta-analysi':4,46,837,894,954,1018,2367,2419,3047,3056,3352 'meta-analyst':813,1868 'metadata':591,3361 'methodolog':967,2795 'mh':712 'mh.text':718 'might':888 'min':3486 'minimum':217,276,1054 'minut':1964,2061 'miss':889,3390 'mm':2959 'modal':900,1193,1553,2224,2406,3145 'mode':51,84,88,93,174,228,283,295,328,329,1429,1434,1435,1954,1967,2022,2057,2067,2185,2499,2890,3004,3018,3189 'monitor':1548 'most-publish':2753 'move':2132,2877 'mri':1198,1562,3157 'msk':3184 'multi':787,1861,2335 'multi-sourc':786,1860,2334 'multipl':1787 'must':399,3552,3590 'n':780 'name':95,180,219,364,379,402,471,481,489,582,641,650,756,1362,1364,1375,1612,1803,2095,2859,2902,2906 'nativ':182 'native-languag':181 'natur':736,2714 'ncbi':557,2166 'need':1755,1767,2041,2798,3050,3432 'neighbor':2253 'network':2687 'neuro':257,3183 'never':381,2103,3540,3561,3577 'new':1538,1596,3071,3144 'next':1487,3013 'nich':1179,1218,1502,1577,2759 'nn':1366,1381,2860 'nodul':242,2220,2264,2274,2283,2293,2867 'non':2523 'non-overlap':2522 'none':1336 'ntm':3130 'number':1278,1383,3464,3551 'often':503,1205,1550,1563,2121,3076,3172 'old':1062,1594 'one':222 'ongo':1001 'open':1551 'opportun':1064 'option':264,2671 'orcid':192 'otherwis':3290 'outcom':850,865,2234,2394,3345 'output':766,1313,1457,1942,2435,2641,2812,3447,3560 'overestim':2114,3599 'overlap':960,1043,1911,2524,2697 'overlay':1532,3096 'packag':3323,3499 'page':375,497 'paper':678,969,2088,3405,3410 'parallel':597,627,634,750,811,1781,1793,1826,1840,1863,1873,1958,1969,1989,2011,2351,3465,3495 'parasit':1584,3131 'pars':2213 'parse_pubmed.py':549 'partial':971 'patient':2249 'peer':2804 'per':924,1960,1965,1991,2062,3472 'percutan':3395 'persist':1460 'phase':29,66,288,302,314,320,335,513,781,925,1224,1339,1455,1794,1822,1856,1935,1940,1943,1973,1983,2015,2024,2035,2193,2330,2469,2489,2493,2557,2665,2826,2892,3006,3295 'photon':3153 'photon-count':3152 'pico':1744 'pico/pird':162,1443,1714,3205,3271 'pillar':530,659,668,738,744,767,770,792,808,834,854,897,1016,1183,1321,1509,1820,1878,2713,3548 'pipelin':3300,3315,3328 'pitfal':2073 'plan':466,1453,3265 'pmid':374,575,600,1807,1816,3369 'pmid1':611,686 'pmid2':612,687 'pmid20':688 'point':58 'pool':950,2380 'popul':1580,2243,3119,3121 'population-specif':1579,3120 'posit':3430 'post':992,3302 'post-scout':3301 'potenti':2760 'power/sample-size':1775 'practic':3084 'pre':27,64,3298,3313,3502,3531 'pre-propos':3297,3312,3501 'pre-protocol':26,63 'pre-registr':3530 'preclin':2184 'precondit':2954 'predict':244,2266,2276 'prefer':156,196,260 'prefix':2855 'preliminari':1445,1764,3459 'prepar':3317 'preprint':988,994,1006,1662,1692,1914,2415,2448,3247 'press':979,1689,3245 'prevent':443 'primari':824,846,1009,1046,1597,1892,1998,2108,2378,2442,2531,3341 'print':727 'prior':1265,1322,1513 'prisma':3412,3453 'proceed':3587 'professor':13,94,102,124,138,177,179,286,332,423,434,524,1247,1254,1319,1361,1370,1372,1385,1522,1568,1693,1961,1966,2076,2129,2509,2695,2703,2711,2857,2879,2921,3327,3402,3497,3506,3579 'professor-first':12,101,176,285,331 'profil':15,97,185,223,458,478,486,515,749,1824,1847,1852,2052,2070 'prognosi':863 'prognost':262,1053,1158,1673,2238,2295,2598,3228 'progress':1665 'project':1346,1497,1751,1753,2836,2852 'prolif':2725 'propos':437,917,2255,2526,2939,3299,3314,3322,3503 'prospero':77,1077,1082,1109,1328,1405,1676,1709,1740,1917,1950,2002,2209,2425,2447,2482,2651,2840,2886,2944,2955,2966,3233,3529,3571 'prospero-readi':1404,1949,2839,2885 'prospero-registr':2208 'protocol':28,65,1003,1103,1681,1741,2427,3238 'provid':99,370,453,500,1851 'pt':840,844,2370,2374,3355,3359 'public':573,1257,1695,3542 'publish':2755,2791,3288 'pubm':193,226,358,469,532,546,752,801,818,885,1000,1123,1641,1691,1715,1994,2044,2112,2356,2377,2720,3214,3272,3555,3569,3597,3612 'pulmonari':2292 'python3':585,613,689 'qualiti':1600,2529,3187 'queri':890,1015,2402,2417 'question':21,105,146,161,234,279,1704,2204,3202 'question-first':145 'rad':2303 'radiolog':251,482,491,757,2551,3040,3060,3090 'radiologist':2271 'radiom':2280 'rank':33,70,905,1117,1227,1229,1314,1317,1468,1937,2560,2566,2642,3009 'rapid':2342 'rate':550,928,2141,2149,2159 'rate-limit':2148 'rather':3623 'raw':1122,1143,1164,1172,1326,1925,2111,2444,3596,3607 're':3423,3429 're-assess':3422 're-posit':3428 'readi':1406,1738,1747,1951,2841,2887,3320 'readm':164,322,1342,1349,1472,1711,1742,1952,2808,2829,2842,3268 'readme.md':1402,2882,3457 'ready-to-propos':3319 'real':1129 'realist':1119,1142,1153,1166,1327,1668,1927,2446,2593,3223,3589,3609 'recenc':1057 'recent':1008,1071,1272,2774,3036,3444 'recommend':1486,2816,3012,3291 'record':1696 'recruit':3267 'redirect':2100 'refer':1425,2701 'references/project_readme_template.md':2895 'refin':2199,3203 'region':1578 'regist':1100,1679,2426,3237 'registered-but-unpublish':1099 'registr':1710,2210,2945,2956,2967,3532 'relat':3139 'relationship':351,431 'reli':510,2104 'reliabl':564,1630 'renam':2920 'repeat':655 'report':1162,1924,3605 'request':427 'requir':218,277,463,1774 'research':44,160,529,737,1703 'researchg':485,490 'resolv':341,361,1613 'respons':1200,1543 'result':906,1494,2750,3010,3386 'retir':216 'retmax':3339 'retract':3409 'retri':2154 'return':566,902,2086,3613 'reveal':735 'revers':2721 'review':843,892,968,1023,2373,2404,2800,2973,2990,3262,3358,3524 'risk':1115,1161 'role':266,3519 'root':700 'root.findall':709 'row':2915 'rr/md':2241 'rsna':3075 'rule':149,1521 'run':631,809,1882,2347,2484,3310 'satur':1045,1185,1206,1511,1546 'save':1466,1493,3008 'scaffold':1343,1757,1945,2830 'scan':1482,2064,2333,3034 'scholar':191,477,483,748,758,803,931,940,1846,1907,2051,2768,3216 'scholarwork':187,493 'scope':253,918,959,1042,1910,1979,2197,2211,2527 'scope-overlap':958,1909 'scopus':462,1855,2106 'scopus/sciencedirect':502,2097 'score':1235,1236 'scout':3,37,1728,3303,3306,3388 'screen':78,2246,2317,2968,2979,3182,3416,3523 'script':547,2169 'se/sp':2237 'search':136,346,359,387,398,472,580,625,639,648,833,853,879,887,935,987,1081,1446,1677,1716,1745,1760,1765,1899,2047,2363,2384,2722,2733,2962,3035,3054,3083,3234,3274,3331,3342,3349,3460,3479,3570 'search-lit':1759 'searchabl':2207 'section':3461 'secur':2974,3534 'see':2894 'select':85,2321,2340 'self':2779,2949,3283 'self-l':2778,2948,3282 'semant':870,933,1897,2399 'semanticsearch':942 'senior':2678,2782 'sensit':857,2388 'separ':1393,3392 'sequenti':1382,1938,1946,1975 'server':1024,2176,2181,2422 'set':2247 'share':325,1004 'signal':92 'signific':1595 'similar':934 'sinc':1269,1599 'site':1085,2429 'skill':38,60,1414 'skill-ma-scout' 'skip':448,1854 'skip/niche':3445 'skku':188,494 'slug':1368,1389 'solo':267,1433,2821 'solo-mod':1432 'solo-vi':2820 'sourc':788,1631,1862,1933,2008,2110,2336,3221 'source-aperivue' 'span':3169 'specialti':203 'specif':373,620,859,1192,1207,1581,1706,1838,2244,2390,2729,3020,3030,3122,3191 'spine':3185 'sr/ma':1002 'srs/mas':883 'start':57,382,2941 'state':3620 'step':570,588,602,616,661,745,1488,2501,2940,2951,3014,3329 'still':2797 'straightforward':2796 'strategi':1717,1746,1783,1970,2684,2718,2776,3275 'structur':567,829 'studi':825,847,1010,1132,1281,1598,1893,2379,2532,3101,3515 'submiss':2995 'subpopul':1208 'subspecialti':252,2918,3091,3167,3171 'subtop':1180,1526 'subtyp':1211 'suffici':3252 'suggest':1490,3027 'summar':1479 'summari':606,768,1458,2437,3001 'supervis':1428 'supervisor':140,274,1438,2899,2913,2938 'supervisor-area':2912 'supervisor-first':139 'supervisor-match':273 'supervisor-propos':2937 'suppli':118 'supporting/updating':1572 'suspect':658 'synonym':3338 'synthesi':3151 'sys':692 'sys.stdin.read':702 'system':1499 'systemat':842,891,1022,2372,2403,3357 't-phase':300,1971,1981,2013,2033,2191,2328,2467,2555,2663,2824 'tabl':769,1316,1449,1470,2438,2644,2817,3452 'target':1447,1720,3276 'tb':1586,3129 'team':2924 'technolog':3142 'templat':323,1403,1410,2883,2888,2947 'tempor':1588 'term':664,721,731,734,775,1818 'terminolog':169,948,1658,2412 'thorac':205,255,3178 'time':2953 'timelin':1454,1724,2934,2946,3280 'titl':3372 'title-bas':3371 'tomograph':742 'top':676,732,1814,2054,2751 'top-cit':675 'topic':7,19,34,49,71,112,115,121,126,144,231,298,436,619,798,898,974,1028,1089,1149,1191,1226,1232,1295,1315,1318,1353,1367,1388,1421,1469,1481,1484,1503,1535,1605,1734,1837,2063,2180,2188,2195,2364,2385,2405,2418,2433,2559,2562,2568,2574,2643,2717,2730,2734,2793,2832,2850,2854,2861,2960,3015,3026,3103,3107,3168,3196,3309,3343,3350,3400,3475,3488,3536 'topic-agent-skills' 'topic-biostatistics' 'topic-claude-code' 'topic-claude-skills' 'topic-clinical-research' 'topic-diagnostic-accuracy' 'topic-first':18,114,230,297,2187,2561,2573,2831,2849,3195 'topic-irb-protocol' 'topic-literature-review' 'topic-manuscript' 'topic-medical-ai' 'topic-medical-research' 'topic-meta-analysis' 'topic-specif':618,1836 'total':572,1480,1805,1903,1953,2056 'treatment':1199,1542,1547 'trend':107,3033 'tri':1092,1515 'triag':3374,3451,3481 'truli':2521 'tune':440 'two':50,3170 'type':259,2235 'typic':158,2030 'unavail':3619 'underscor':1392 'underscore-separ':1391 'underserv':1564 'uniform':2540 'unique/niche':1525 'univers':495 'unpublish':1102 'updat':1063,1589,2586,3069,3425,3440,3458 'upgrad':2304 'urgenc':3527 'url':98,186,224,459,499,1853 'use':531,538,800,812,874,936,963,982,1212,1939,2093,2175,2410,2490,2853,2916,3591 'user':90,133,153,349,414,417,429,451,1464,1492,1627,1976,2201,2259,2320,2541,2549,2599,2675,2737,2788,2811,2901,2905,2926,3023,3248,3285,3585 'util':519,535,541,821,1617,1644,1800,1812,1833,1887,2158,2744,3335,3558 'valid':1649 'variant':1439,2267,2278,2286,2296,2307 'variat':2258,3128 'verdict':1330 'verif':2139 'verifi':3567 'via':628,1083,1614,1640,1650,3213,3568 'viabl':796,1231,1352,1483,1607,2028,2477,2567,2786,2822,3200,3308 'vs':1197,1201,1561,2270,2637,3394 'wait':2151 'want':2676 'warrant':2536 'wave':3473 'webfetch':544,1093,1849 'websearch':479,487,754,1084,1844,1920,2049,2428 'week':2965,2971,2978,2984,2992,2997 'weight':1245,2564,2570,2578 'well':3100 'well-studi':3099 'whether':134,418,1293 'whose':2712 'within':1369,1384,3385 'without':2138,3028 'work':474,1357,1476,2707,2846 'workflow':282,334,2190 'write':3482 'wrong':445 'wrong-institut':444 'www.crd.york.ac.uk':1095 'www.crd.york.ac.uk/prospero/#searchadvanced':1094 'x':2118,3603 'xml':673 'xml.etree.elementtree':693 'year':1061,1593,1639 'yet':997 'yr':1268 'yyyi':2958 'yyyy-mm':2957 '가능':2636 '가능성':2612,2625 '가이드라인':2610 '경쟁':2458 '관계':2627 '기존':2440,2583,2626,2647 '기회':2587 '논문':777 '높음':2602 '단독':2634 '대표':772 '맞는가':2606 '변경':2611 '변형':2439 '분야':2622 '분야와':2605 '사용자의':2603 '선택적':2661 '수':778 '순위':2645 '없음':2585,2660 '영역':771 '용이':2630 '의사결정':2614 '임상':2613 '전문':2604 '전문가':2623,2638 '접근':2624,2629 '종합':2656 '주제':2646 '중간':2609,2620,2633 '직결':2615 '진행':2635 '최고':2582,2591 '최적':2466,2662 '추정':776,2649 '키워드':773 '판정':2450 '포화':2588 '필수':2640 '필요':2655 '해당':2621 '해석':2639","prices":[{"id":"dd3931b6-80d8-4126-9315-eb37a5406829","listingId":"3518fce5-df88-4f35-a30b-aec1fb0cccad","amountUsd":"0","unit":"free","nativeCurrency":null,"nativeAmount":null,"chain":null,"payTo":null,"paymentMethod":"skill-free","isPrimary":true,"details":{"org":"Aperivue","category":"medsci-skills","install_from":"skills.sh"},"createdAt":"2026-05-13T12:57:45.393Z"}],"sources":[{"listingId":"3518fce5-df88-4f35-a30b-aec1fb0cccad","source":"github","sourceId":"Aperivue/medsci-skills/ma-scout","sourceUrl":"https://github.com/Aperivue/medsci-skills/tree/main/skills/ma-scout","isPrimary":false,"firstSeenAt":"2026-05-13T12:57:45.393Z","lastSeenAt":"2026-05-18T18:56:30.338Z"}],"details":{"listingId":"3518fce5-df88-4f35-a30b-aec1fb0cccad","quickStartSnippet":null,"exampleRequest":null,"exampleResponse":null,"schema":null,"openapiUrl":null,"agentsTxtUrl":null,"citations":[],"useCases":[],"bestFor":[],"notFor":[],"kindDetails":{"org":"Aperivue","slug":"ma-scout","github":{"repo":"Aperivue/medsci-skills","stars":98,"topics":["agent-skills","biostatistics","claude-code","claude-skills","clinical-research","diagnostic-accuracy","irb-protocol","literature-review","manuscript","medical-ai","medical-research","meta-analysis","physician-researcher","prisma","pubmed","radiology","reporting-guidelines","strobe","systematic-review","tripod-ai"],"license":"other","html_url":"https://github.com/Aperivue/medsci-skills","pushed_at":"2026-05-17T20:50:52Z","description":"Claude Code skills for medical research — literature search, reporting guidelines, statistical analysis, publication figures. Built by a physician-researcher, tested on real publications. MIT licensed.","skill_md_sha":"8722ff5386fb0f5662330382ce2adfa67aea0679","skill_md_path":"skills/ma-scout/SKILL.md","default_branch":"main","skill_tree_url":"https://github.com/Aperivue/medsci-skills/tree/main/skills/ma-scout"},"layout":"multi","source":"github","category":"medsci-skills","frontmatter":{"name":"ma-scout","description":"Meta-analysis topic discovery and feasibility assessment. Professor-first (profile → gap) or Topic-first (question → gap → co-author). Pre-protocol phase from idea to ranked topic list."},"skills_sh_url":"https://skills.sh/Aperivue/medsci-skills/ma-scout"},"updatedAt":"2026-05-18T18:56:30.338Z"}}