Medical Review Topic Scout

Find review topics that are not too empty, not too saturated, and easier to justify with PubMed evidence.

medical-review-topic-scout is a local Codex skill plus a standalone Python scanner for biomedical literature review topic scouting. It turns a rough pair such as gastric cancer + cancer-associated fibroblasts into candidate review angles, PubMed search strings, recent-review competition checks, and a Chinese-style topic report.

Highlights

• Evidence-first topic scouting: turn rough disease/mechanism ideas into reproducible PubMed search checks.
• Competition-aware scoring: compare total evidence, recent reviews, and stricter title-level direct reviews.
• Report-ready outputs: export search logs, sample papers, viability labels, and Chinese topic report structure.
• Local and lightweight: works as a Python script or as a local Codex skill, with no API key required for light use.

What it helps you answer

• Is this review topic already saturated?
• Are there enough recent primary studies to support a manuscript?
• Are there direct competing reviews in the last 2-3 years?
• Which narrower mechanism, biomarker, intervention, omics, or translational angle is more writable?
• What exact PubMed query should be recorded in the topic report?

Workflow

flowchart LR
    A["Input pair<br/>disease + mechanism / biomarker / intervention"] --> B["Generate candidate angles<br/>broad, mechanistic, translational"]
    B --> C["Build PubMed queries<br/>synonyms + MeSH + date windows"]
    C --> D["Scan evidence volume<br/>5-year papers, reviews, title-level reviews"]
    D --> E["Score topic viability<br/>literature sufficiency + competition safety"]
    E --> F["Write Chinese topic report<br/>recommended title, gap, outline, search log"]

Loading

Repository contents

Path	Purpose
SKILL.md	Codex skill instructions and report workflow
scripts/pubmed_topic_scan.py	PubMed E-utilities scanner for counts, direct-review checks, query translation capture, scoring, JSON/CSV export, synonym blocks, and MeSH terms
references/report-template.md	Copyable Chinese topic report template
references/release-process.md	Release note and version-maintenance guidance
agents/openai.yaml	UI metadata for the skill

Quick start: run the PubMed scanner

Requirements:

• Python 3.9+
• Internet access to NCBI PubMed E-utilities
• No API key is required for light use

git clone https://github.com/sushuqiong/medical-review-topic-scout.git
cd medical-review-topic-scout

python .\scripts\pubmed_topic_scan.py `
  --direction "gut microbiome|Alzheimer disease"

Multiple candidate directions can be scanned at once:

python .\scripts\pubmed_topic_scan.py `
  --direction "gut microbiome|Alzheimer disease" `
  --direction "tau/phosphorylated tau|gut microbiome/gut microbiota|Alzheimer disease/Alzheimer's disease" `
  --csv topic_scan.csv

The script reports:

• total PubMed hits in the last 5 years
• all review hits
• recent 3-year review hits
• stricter title-level recent review hits
• a rough viability score
• PubMed QueryTranslation
• sample recent/direct reviews and latest papers

Use as a Codex skill

Clone or copy this repository into your local Codex skills directory, then invoke:

Use $medical-review-topic-scout 我想写一篇 胃癌 和 肿瘤相关成纤维细胞 相关的综述，帮我找个文献量够、直接竞争少的选题。

Example tasks:

Use $medical-review-topic-scout 帮我评估“结直肠癌 + ferroptosis”是否适合写综述，并给出3个更窄的机制方向。

Use $medical-review-topic-scout 我想写“糖尿病肾病 + 单细胞测序”方向，请帮我查PubMed竞争综述并生成选题报告。

How directions are written

Use | to separate concept blocks and / to separate synonyms inside a block.

disease synonym 1/disease synonym 2/mesh:MeSH Term|mechanism synonym 1/mechanism synonym 2

Examples:

gastric cancer/stomach neoplasms/mesh:Stomach Neoplasms|cancer-associated fibroblasts/CAF

diabetic kidney disease/diabetic nephropathy/mesh:Diabetic Nephropathies|single-cell RNA sequencing/scRNA-seq

Saturation labels

Label	Meaning
严重饱和	Multiple title-level recent direct reviews; usually avoid unless doing a clear update
饱和	Many papers plus many recent reviews; differentiation is difficult
近期有综述	A nearby review exists recently; scope needs careful repositioning
刚被占/需差异化	One recent direct review appears; possible but risky
有空间	Enough evidence and limited direct review competition
空白/有空间	No direct reviews and enough recent literature to support writing
略窄	Literature volume may be too small unless adjacent evidence is strong

Example output

| Direction | 5-year count | Reviews all | Reviews recent 3y | Title-review recent 3y | Score | Label |
|---|---:|---:|---:|---:|---:|---|
| gut microbiome/gut microbiota/... x Alzheimer disease/... | ... | ... | ... | ... | ... | ... |

## gut microbiome/gut microbiota/... x Alzheimer disease/...
- 5-year query: `...`
- Review query: `...`
- PubMed query translation: `...`
- Recent/direct review samples:
  - PMID ...: ...

Recommended report structure

The skill writes the final answer in Chinese by default:

1. Topic landscape table
2. Recommended title
3. Evidence volume support
4. Direct-review competition check
5. Novelty and differentiation statement
6. Suggested manuscript outline
7. PubMed search log
8. Backup directions
9. Risks and next steps

Notes for medical research use

• PubMed counts are search-dependent and should be treated as reproducible scouting evidence, not final proof of novelty.
• For formal manuscript planning, verify important topics in Web of Science, Scopus, Embase, or another database available to your institution.
• Do not claim a topic has “never been written” unless reproducible searches support that conclusion.
• Avoid entering patient identifiers or protected health information into search prompts.
• This project supports literature topic selection and does not provide clinical diagnosis or treatment advice.

Project maintenance

• CHANGELOG.md
• references/release-process.md
• .github/release-template.md

Keywords

PubMed · biomedical literature · review topic scouting · evidence synthesis · medical research · systematic review · AI for science