Source: https://github.com/aipoch/medical-research-skills
Output Format
Output must strictly follow this structured report format to ensure users receive directly readable decision-ready content rather than a simple text description.
1. Hotspot Overview Table
Display all identified hotspot topics and their popularity metrics in tabular form.
| Hotspot Topic |
Popularity Index |
Trend Direction |
Representative Papers |
Active Years |
| [Topic 1] |
★★★★★ |
Rising |
45 papers |
2023-2026 |
| [Topic 2] |
★★★★ |
Stable |
28 papers |
2022-2026 |
| [Topic 3] |
★★★ |
Declining |
12 papers |
2020-2024 |
- Popularity Index: Composite score based on publication volume, citation frequency, and top-tier journal proportion (max ★★★★★).
- Trend Direction: Recent 3-year publication trend (Rising / Stable / Declining).
- Representative Papers: Number of core papers matching the hotspot topic.
- Active Years: Year range with sustained output for the hotspot.
2. Hotspot Detail Analysis
Each hotspot is expanded independently, including popularity rating, trend description, sub-direction composition, and key papers.
Hotspot 1: [Topic Name]
- Popularity Index: ★★★★★ (Very Hot)
- Trend: Rising publication volume over the past 3 years
- Core Research Directions:
- Sub-direction A (40%)
- Sub-direction B (35%)
- Key Papers:
| Paper |
Journal |
Year |
Citations |
Evidence Level |
| [Title] |
Nature |
2025 |
230 |
High |
| [Title] |
Cell Rep |
2024 |
98 |
Medium |
| [Title] |
Front Immunol |
2024 |
15 |
Low |
- Evidence Level: High (top-tier/highly cited), Medium (mainstream/moderate citations), Low (lower-tier/few citations).
Hotspot 2: [Topic Name]
...
3. Research Gap Identification
Analyze under-explored areas in current literature.
| Research Gap |
Potential Value |
Feasibility |
Reason |
| [Gap 1 - underexplored direction] |
High |
Medium |
Few than 10 papers, but clear clinical demand |
| [Gap 2] |
Medium |
High |
Mature tools available, but not yet applied in this field |
- Potential Value: High / Medium / Low, based on unmet clinical or basic research needs.
- Feasibility: High / Medium / Low, based on technical maturity, research barriers, and execution difficulty.
4. Recommended Entry Directions
Based on the preceding analysis, provide concrete actionable research entry suggestions.
| Priority |
Recommended Direction |
Reason |
Expected Output |
| 1 |
[Direction A] |
High popularity + existing gap + good feasibility |
1 review / 1 experimental design |
| 2 |
[Direction B] |
Emerging hotspot + low competition |
1-2 research papers |
| 3 |
[Direction C] |
Niche but high clinical value |
Case series / methodology paper |
Research Hotspot Analysis
When to Use
- The user provides a disease name, target, technical roadmap, or research topic, and wants to quickly see current research hotspots.
- The user needs to cluster recent literature by keywords and topics to find directions worth deeper exploration.
- The user wants a Markdown hotspot analysis report with representative literature for topic selection or review writing.
When Not to Use
- Do not use this skill when the user only needs single-paper retrieval or a simple reference list.
- If there is no clear disease, topic, or search scope, do not start clustering immediately — first ask the user to clarify topic boundaries.
- If the environment cannot access the scripts or retrieval data this skill depends on, do not fabricate hotspot results.
Required Inputs
| Field |
Required |
Format/Source |
Example |
If Missing |
topic |
Yes |
Text |
lung cancer immunotherapy |
Stop and request topic |
time_range |
No |
Time range |
last 5 years |
Default to recent literature |
focus |
No |
Text |
mechanism, clinical translation |
Default to comprehensive hotspot output |
Workflow
- Use
search_pubmed from scripts/analysis_ops.py to search relevant literature and obtain PMIDs and basic metadata.
- Run
word_frequency on the returned medline_texts to count high-frequency keywords or MeSH terms.
- Combine with hotspot prompts in
references/prompt_templates.md to cluster high-frequency keywords into 3-6 hotspot topics.
- Use
match_keywords to map representative literature to each topic, avoiding mismatches between topics and evidence.
- For each topic, call
sort_by_jif_and_select to choose representative literature, then use fetchPMCArticleDetails or fetchPubmedArticleDetails to supplement details.
- Output a Markdown report with at least: research overview, hotspot topics, representative keywords per topic, representative literature, and follow-up suggestions.
Output Contract
- Primary output: A Markdown hotspot analysis report.
- Required fields:
topic overview, hotspot topics, supporting papers, next-step suggestions.
- Recommend at least 2-3 representative papers per hotspot, with explanation of why the topic qualifies as a hotspot.
- If retrieval coverage is insufficient, must explicitly mark as
PARTIAL.
Failure Handling
- Too few literature search results: First broaden time range or relax keywords, then explain coverage gaps.
- Unstable keyword clustering: Show high-frequency keywords and indicate clustering is candidate-only — do not force conclusions.
- Representative literature lacks usable details: Keep PMID and title, mark as pending.
User Checkpoints
- Before starting a broad search, confirm topic boundaries and time range.
- Before outputting the final hotspot report, if cluster topics are clearly ambiguous, send candidate topics to user for confirmation.
Tools
fetchPMCArticleDetails: Get article details.fetchPubmedArticleDetails: Get PubMed details.
Scripts
scripts/analysis_ops.py: Contains helper functions for PubMed search, frequency analysis, keyword matching, and result formatting.
References
references/prompt_templates.md: Contains the system prompts for LLM analysis.
Input Validation
This skill accepts requests that match the documented purpose of research-hotspot-analysis and include enough context to complete the workflow safely.
Quick Validation
- Check that
scripts/analysis_ops.py exists and can perform at least the three core steps: search, word frequency, and matching.
- Check that the final report contains hotspot topics with corresponding representative literature, not just a keyword list.
- Check that each hotspot topic has clear evidence sources to support it.
1---2name: research-hotspot-analysis3description: Analyze research hotspots for a disease or topic and recommend representative literature. Use when users need to identify trending directions, topic clusters, or generate hotspot review reports. Input is a disease name or research topic; output is a structured hotspot analysis report and representative literature list.4license: MIT5---6> **Source**: [https://github.com/aipoch/medical-research-skills](https://github.com/aipoch/medical-research-skills)
7
8## Output Format
9
10Output must strictly follow this structured report format to ensure users receive directly readable decision-ready content rather than a simple text description.
11
12### 1. Hotspot Overview Table
13
14Display all identified hotspot topics and their popularity metrics in tabular form.
15
16| Hotspot Topic | Popularity Index | Trend Direction | Representative Papers | Active Years |
17|---------|:-------:|:--------:|:-----------:|:--------:|
18| [Topic 1] | ★★★★★ | Rising | 45 papers | 2023-2026 |
19| [Topic 2] | ★★★★ | Stable | 28 papers | 2022-2026 |
20| [Topic 3] | ★★★ | Declining | 12 papers | 2020-2024 |
21
22- Popularity Index: Composite score based on publication volume, citation frequency, and top-tier journal proportion (max ★★★★★).
23- Trend Direction: Recent 3-year publication trend (Rising / Stable / Declining).
24- Representative Papers: Number of core papers matching the hotspot topic.
25- Active Years: Year range with sustained output for the hotspot.
26
27### 2. Hotspot Detail Analysis
28
29Each hotspot is expanded independently, including popularity rating, trend description, sub-direction composition, and key papers.
30
31#### Hotspot 1: [Topic Name]
32- **Popularity Index**: ★★★★★ (Very Hot)
33- **Trend**: Rising publication volume over the past 3 years
34- **Core Research Directions**:
35 - Sub-direction A (40%)
36 - Sub-direction B (35%)
37- **Key Papers**:
38
39| Paper | Journal | Year | Citations | Evidence Level |
40|------|:----:|:----:|:----:|:--------:|
41| [Title] | Nature | 2025 | 230 | High |
42| [Title] | Cell Rep | 2024 | 98 | Medium |
43| [Title] | Front Immunol | 2024 | 15 | Low |
44
45- **Evidence Level**: High (top-tier/highly cited), Medium (mainstream/moderate citations), Low (lower-tier/few citations).
46
47#### Hotspot 2: [Topic Name]
48...
49
50### 3. Research Gap Identification
51
52Analyze under-explored areas in current literature.
53
54| Research Gap | Potential Value | Feasibility | Reason |
55|---------|:--------:|:------:|:--------:|
56| [Gap 1 - underexplored direction] | High | Medium | Few than 10 papers, but clear clinical demand |
57| [Gap 2] | Medium | High | Mature tools available, but not yet applied in this field |
58
59- Potential Value: High / Medium / Low, based on unmet clinical or basic research needs.
60- Feasibility: High / Medium / Low, based on technical maturity, research barriers, and execution difficulty.
61
62### 4. Recommended Entry Directions
63
64Based on the preceding analysis, provide concrete actionable research entry suggestions.
65
66| Priority | Recommended Direction | Reason | Expected Output |
67|:------:|:--------|:-----|:---------|
68| 1 | [Direction A] | High popularity + existing gap + good feasibility | 1 review / 1 experimental design |
69| 2 | [Direction B] | Emerging hotspot + low competition | 1-2 research papers |
70| 3 | [Direction C] | Niche but high clinical value | Case series / methodology paper |
71
72---
73
74# Research Hotspot Analysis
75
76## When to Use
77
78- The user provides a disease name, target, technical roadmap, or research topic, and wants to quickly see current research hotspots.
79- The user needs to cluster recent literature by keywords and topics to find directions worth deeper exploration.
80- The user wants a Markdown hotspot analysis report with representative literature for topic selection or review writing.
81
82## When Not to Use
83
84- Do not use this skill when the user only needs single-paper retrieval or a simple reference list.
85- If there is no clear disease, topic, or search scope, do not start clustering immediately — first ask the user to clarify topic boundaries.
86- If the environment cannot access the scripts or retrieval data this skill depends on, do not fabricate hotspot results.
87
88## Required Inputs
89
90| Field | Required | Format/Source | Example | If Missing |
91|---|---|---|---|---|
92| `topic` | Yes | Text | `lung cancer immunotherapy` | Stop and request topic |
93| `time_range` | No | Time range | `last 5 years` | Default to recent literature |
94| `focus` | No | Text | `mechanism`, `clinical translation` | Default to comprehensive hotspot output |
95
96## Workflow
97
981. Use `search_pubmed` from `scripts/analysis_ops.py` to search relevant literature and obtain PMIDs and basic metadata.
992. Run `word_frequency` on the returned `medline_texts` to count high-frequency keywords or MeSH terms.
1003. Combine with hotspot prompts in `references/prompt_templates.md` to cluster high-frequency keywords into 3-6 hotspot topics.
1014. Use `match_keywords` to map representative literature to each topic, avoiding mismatches between topics and evidence.
1025. For each topic, call `sort_by_jif_and_select` to choose representative literature, then use `fetchPMCArticleDetails` or `fetchPubmedArticleDetails` to supplement details.
1036. Output a Markdown report with at least: research overview, hotspot topics, representative keywords per topic, representative literature, and follow-up suggestions.
104
105## Output Contract
106
107- Primary output: A Markdown hotspot analysis report.
108- Required fields: `topic overview`, `hotspot topics`, `supporting papers`, `next-step suggestions`.
109- Recommend at least 2-3 representative papers per hotspot, with explanation of why the topic qualifies as a hotspot.
110- If retrieval coverage is insufficient, must explicitly mark as `PARTIAL`.
111
112## Failure Handling
113
114- Too few literature search results: First broaden time range or relax keywords, then explain coverage gaps.
115- Unstable keyword clustering: Show high-frequency keywords and indicate clustering is candidate-only — do not force conclusions.
116- Representative literature lacks usable details: Keep PMID and title, mark as pending.
117
118## User Checkpoints
119
120- Before starting a broad search, confirm topic boundaries and time range.
121- Before outputting the final hotspot report, if cluster topics are clearly ambiguous, send candidate topics to user for confirmation.
122
123## Tools
124* `fetchPMCArticleDetails`: Get article details.
125* `fetchPubmedArticleDetails`: Get PubMed details.
126
127## Scripts
128* `scripts/analysis_ops.py`: Contains helper functions for PubMed search, frequency analysis, keyword matching, and result formatting.
129
130## References
131* `references/prompt_templates.md`: Contains the system prompts for LLM analysis.
132
133## Input Validation
134
135This skill accepts requests that match the documented purpose of `research-hotspot-analysis` and include enough context to complete the workflow safely.
136
137## Quick Validation
138
139- Check that `scripts/analysis_ops.py` exists and can perform at least the three core steps: search, word frequency, and matching.
140- Check that the final report contains hotspot topics with corresponding representative literature, not just a keyword list.
141- Check that each hotspot topic has clear evidence sources to support it.
gabrielmoreira