skills-collection

Skills Collection — Medical AI PhD Research Workflows

A curated collection of 50 Claude Code skills organized by research workflow stage, tailored for AI and Medical AI PhD researchers.

Documentation

File	Description
PLATFORM_GUIDE.en.md	Which skills work in Claude Code vs Claude.ai
PLATFORM_GUIDE.zh.md	平台兼容性指南（中文）
recommended_skills_for_medical_ai_phd.en.md	Recommended skills by research stage (English)
recommended_skills_for_medical_ai_phd.zh.md	按研究阶段推荐的 skills（中文）

Plugins

Plugin	Skills	Description
literature-and-topic-selection	12	Topic ideation, literature review, academic databases
data-preparation-and-processing	12	Medical imaging, EHR, physiological signals, data processing
model-development-and-experiments	11	Deep learning, GNNs, explainability, time series
results-analysis-and-visualization	6	Plotting, statistical analysis, interactive visualization
paper-writing-and-submission	10	Academic writing, slides, posters, peer review

Installation

Step 1 — Add this repository as a marketplace

/plugin marketplace add youyinnn/skills-collection

Step 2 — Install the plugins you need

/plugin install literature-and-topic-selection@skills-collection
/plugin install data-preparation-and-processing@skills-collection
/plugin install model-development-and-experiments@skills-collection
/plugin install results-analysis-and-visualization@skills-collection
/plugin install paper-writing-and-submission@skills-collection

You can install only the plugins relevant to your current work.

Recommended minimal install for Medical AI PhD

/plugin install literature-and-topic-selection@skills-collection
/plugin install data-preparation-and-processing@skills-collection
/plugin install model-development-and-experiments@skills-collection
/plugin install results-analysis-and-visualization@skills-collection
/plugin install paper-writing-and-submission@skills-collection

Skills by Stage

Stage 1 — Literature & Topic Selection

Skill	Purpose
scientific-brainstorming	Research ideation, finding gaps, cross-disciplinary exploration
hypothesis-generation	Derive testable hypotheses from data or literature
literature-review	Systematic review across PubMed / arXiv / bioRxiv
scientific-critical-thinking	Evaluate study design, identify bias, evidence grading (GRADE)
citation-management	Reference management workflows
pyzotero	Programmatic Zotero library management
arxiv-database	Search arXiv preprints (CS / AI / Statistics)
pubmed-database	PubMed REST API with Boolean / MeSH queries
biorxiv-database	Search bioRxiv life science preprints
openalex-database	Query 240M+ scholarly works, citation analysis
bgpt-paper-search	Extract structured experiment data from full-text papers
research-grants	NSF / NIH / DARPA grant writing

Stage 2 — Data Preparation & Processing

Skill	Purpose
pydicom	Read/write DICOM files (CT / MRI / X-Ray / Ultrasound)
pathml	Computational pathology WSI analysis, 160+ formats
histolab	Lightweight H&E tile extraction
imaging-data-commons	Access NIH public medical imaging datasets
pyhealth	Medical AI toolkit: MIMIC-III/IV, eICU, OMOP
clinicaltrials-database	Query ClinicalTrials.gov
clinvar-database	Query ClinVar genetic variants database
neurokit2	Physiological signal processing: ECG / EEG / EDA / PPG
polars	Fast in-memory dataframe processing
dask	Distributed computing for larger-than-RAM workflows
exploratory-data-analysis	EDA reports for scientific data formats
statistical-analysis	Statistical testing and modeling

Stage 3 — Model Development & Experiments

Skill	Purpose
pytorch-lightning	Structured PyTorch training with logging and checkpointing
transformers	HuggingFace transformers for NLP and vision tasks
torch-geometric	Graph neural networks for molecular / biomedical data
scikit-learn	Classical ML: classification, regression, clustering
pymc	Bayesian statistical modeling
shap	Model explainability via Shapley values
umap-learn	Dimensionality reduction and embedding visualization
aeon	Time series classification and regression
networkx	Graph analysis and network science
primekg	Biomedical knowledge graph for drug discovery
stable-baselines3	Reinforcement learning algorithms

Stage 4 — Results Analysis & Visualization