medsci-modeling plugin
Clinical AI model research engineering: choose a paper-grounded architecture, scaffold a reproducible PyTorch training repo, validate the model's split and validation design, document it (Model Card / Datasheet), and evaluate models or LLMs/MLLMs on clinical tasks. Integrates MONAI / nnU-Net, never replaces them.
How to install the medsci-modeling plugin
Run both commands inside Claude Code. The first adds the marketplace once; the second installs the plugin.
/plugin marketplace add Aperivue/medsci-skills
/plugin install medsci-modeling@medsci-skillsWhat's inside medsci-modeling
15 parts read from the plugin's own files.
Skills 15
Claude loads it when the task matches- architecture-zoo
Selects paper-grounded architecture families and generates decision notes through a multi-phase selection process.
- explainability
Perform model assessment tasks including saliency sanity checks, localisation, and attribution framing.
- imaging-data
Profiles medical imaging datasets like DICOM and NIfTI to audit spacing, orientation, and intensity while preventing data leakage during preprocessing.
- mllm-eval
Audits and designs evaluation frameworks for clinical LLM and multimodal LLM tasks including report generation, VQA, and text extraction.
- model-assessment
Validates medical imaging models by auditing data leakage, computing task-specific metrics, and evaluating uncertainty and explainability.
- model-card
Generates standardized documentation for medical-imaging models including Model Cards, Datasheets, and METRIC-informed data-quality assessments.
- model-evaluation
Assess model performance by running assessment phases for task-correct held-out metrics and metric-reporting.
- model-scaffold
Generates a runnable PyTorch training repository for medical imaging tasks with patient-level seed-locked splits and reproducibility guarantees.
- model-selection
Selects medical imaging architectures and vets model artifacts for license, provenance, and benchmark contamination.
- model-sourcing
Selects and gates model dossiers for specific repositories or checkpoints using the model-selection command.
- model-validation
Perform model assessment and validation design through a multi-phase process including split-leakage gates.
- preprocess-imaging
Redirects legacy imaging preprocessing requests to the updated imaging-data endpoint starting from phases 4 through 7.
- profile-imaging
Redirects profile-imaging requests to the imaging-data endpoint to profile datasets and record decisions.
- radiomics-ml
Builds and audits rigorous radiomics and clinical machine learning pipelines using classical learners and nested cross-validation.
- uncertainty-imaging
Assess model uncertainty, out-of-distribution detection, and selective prediction using the model-assessment tool.
Using medsci-modeling in Claude Code
Skills need no command. Claude reads them when your request matches what they describe.
Update, turn off or remove
- Update the plugin (shell)
- claude plugin update medsci-modeling@medsci-skills
- Refresh the marketplace
- /plugin marketplace update medsci-skills
- Turn it off for now
- /plugin disable medsci-modeling@medsci-skills
- Remove it
- /plugin uninstall medsci-modeling@medsci-skills
Skills in medsci-modeling
- imaging-dataAperivue/medsci-skills
Profiles medical imaging datasets like DICOM and NIfTI to audit spacing, orientation, and intensity while preventing data leakage during preprocessing.
26 files378stars - mllm-evalAperivue/medsci-skills
Audits and designs evaluation frameworks for clinical LLM and multimodal LLM tasks including report generation, VQA, and text extraction.
11 files378stars - model-assessmentAperivue/medsci-skills
Validates medical imaging models by auditing data leakage, computing task-specific metrics, and evaluating uncertainty and explainability.
48 files378stars - model-cardAperivue/medsci-skills
Generates standardized documentation for medical-imaging models including Model Cards, Datasheets, and METRIC-informed data-quality assessments.
11 files378stars - model-scaffoldAperivue/medsci-skills
Generates a runnable PyTorch training repository for medical imaging tasks with patient-level seed-locked splits and reproducibility guarantees.
11 files378stars - model-selectionAperivue/medsci-skills
Selects medical imaging architectures and vets model artifacts for license, provenance, and benchmark contamination.
18 files378stars - radiomics-mlAperivue/medsci-skills
Builds and audits rigorous radiomics and clinical machine learning pipelines using classical learners and nested cross-validation.
10 files378stars - architecture-zooAperivue/medsci-skills
Selects paper-grounded architecture families and generates decision notes through a multi-phase selection process.
378stars - explainabilityAperivue/medsci-skills
Perform model assessment tasks including saliency sanity checks, localisation, and attribution framing.
378stars - model-evaluationAperivue/medsci-skills
Assess model performance by running assessment phases for task-correct held-out metrics and metric-reporting.
378stars - model-sourcingAperivue/medsci-skills
Selects and gates model dossiers for specific repositories or checkpoints using the model-selection command.
378stars - model-validationAperivue/medsci-skills
Perform model assessment and validation design through a multi-phase process including split-leakage gates.
378stars - preprocess-imagingAperivue/medsci-skills
Redirects legacy imaging preprocessing requests to the updated imaging-data endpoint starting from phases 4 through 7.
378stars - profile-imagingAperivue/medsci-skills
Redirects profile-imaging requests to the imaging-data endpoint to profile datasets and record decisions.
378stars - uncertainty-imagingAperivue/medsci-skills
Assess model uncertainty, out-of-distribution detection, and selective prediction using the model-assessment tool.
378stars
medsci-modeling plugin questions
How do I install the medsci-modeling plugin?
In Claude Code run /plugin marketplace add Aperivue/medsci-skills, then /plugin install medsci-modeling@medsci-skills.
What does the medsci-modeling plugin include?
medsci-modeling bundles 15 skills.
How do I update or remove the medsci-modeling plugin?
Update it with claude plugin update medsci-modeling@medsci-skills in your shell, or from the Installed tab of /plugin. Remove it with /plugin uninstall medsci-modeling@medsci-skills.