pyhealth

Build clinical/healthcare deep-learning pipelines with PyHealth — loading EHR/signal/imaging datasets (MIMIC-III/IV, eICU, OMOP, SleepEDF, ChestXray14, EHRShot), defining tasks (mortality, readmission, length-of-stay, drug recommendation, sleep staging, ICD coding, EEG events), instantiating models (Transformer, RETAIN, GAMENet, SafeDrug, MICRON, StageNet, AdaCare, CNN/RNN/MLP), training with the PyHealth Trainer, computing clinical metrics, and using medical code utilities (ICD/ATC/NDC/RxNorm lookup and cross-mapping). Use this skill whenever the user mentions PyHealth, MIMIC, eICU, OMOP, EHR modeling, clinical prediction, drug recommendation, sleep staging, medical code mapping, ICD/ATC codes, or any healthcare ML pipeline that fits the dataset → task → model → trainer → metrics pattern, even if "PyHealth" isn't named explicitly.

Quality

93%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Quality

Content

86%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

A strong, well-structured skill body: a runnable canonical pipeline, a high-value gotchas checklist, and a clean reference table pointing to verified bundle files. Minor conciseness loss from description/body trigger duplication and the advisory style note; workflow clarity stops just short of explicit validation feedback loops.

Suggestions

Trim the introductory paragraph and the "When to use this skill" list, since the same triggers already appear in the frontmatter description — keep one authoritative trigger source to recover tokens.

Replace the advisory "A note on style" with a concrete rule (e.g. "Prefer `Trainer` over a custom loop; it handles checkpointing, logging, and best-model selection") to cut philosophy.

Add one explicit validation/feedback step to the pipeline (e.g. after `trainer.train`, confirm `monitor` improved over baseline before relying on the checkpoint) to push workflow clarity to 5.

Dimension	Reasoning	Score
Conciseness	Mostly lean and assumes Claude's competence (no explanation of what a Transformer or DataLoader is), but the introductory paragraph and "When to use this skill" restate triggers already in the description, and the advisory "A note on style" could be trimmed; 4 rather than 5 because not every token earns its place.	4 / 5
Actionability	Fully executable, copy-paste-ready pipeline (~20 lines with real imports and a runnable synthetic-MIMIC URL), concrete `uv` install commands, and specific gotchas with exact constructor args (`ehr_root=` vs `root=`, `cache_dir=`, `monitor=` values by task type) covering the common cases.	5 / 5
Workflow Clarity	The 5-stage pipeline (Dataset → Task → Model → Trainer → Metrics) is clearly sequenced with a numbered, commented code example, and "Critical things to get right" serves as a checklist with a leakage-prevention checkpoint ("Always split by patient"). Falls short of 5 because there is no explicit validate→fix→retry feedback loop, though training validation is implicit via `monitor=`.	4 / 5
Progressive Disclosure	Clear overview in SKILL.md with a well-signaled one-level-deep reference table mapping each topic to a real file (installation/datasets/tasks/models/medcode/examples, all verified present) plus `assets/starter_pipeline.py`; content is appropriately split and easy to navigate.	5 / 5
	Total	18 / 20 Passed

Description

100%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A model description: third-person voice, comprehensive concrete actions, rich natural trigger terms, and explicit what/when guidance with a named-library anchor plus an implicit-trigger escape hatch. No vague fluff or over-claims.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions across the full pipeline — "loading EHR/signal/imaging datasets", "defining tasks", "instantiating models", "training with the PyHealth Trainer", "computing clinical metrics", and "medical code utilities (ICD/ATC/NDC/RxNorm lookup and cross-mapping)" — with comprehensive coverage and no real gaps.	5 / 5
Completeness	Explicitly answers both what ("Build clinical/healthcare deep-learning pipelines with PyHealth ...") and when ("Use this skill whenever the user mentions ..."), with concrete trigger phrases; the "even if PyHealth isn't named explicitly" clause strengthens the when.	5 / 5
Trigger Term Quality	Comprehensive natural-term coverage including synonyms and the canonical library name: "PyHealth, MIMIC, eICU, OMOP, EHR modeling, clinical prediction, drug recommendation, sleep staging, medical code mapping, ICD/ATC codes" — all phrases a user would naturally say.	5 / 5
Distinctiveness Conflict Risk	Clear niche (the PyHealth clinical/healthcare ML library) with distinct, domain-specific triggers; minimal overlap risk with other skills.	5 / 5
	Total	20 / 20 Passed

Validation

100%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository: K-Dense-AI/scientific-agent-skills
Path: skills/pyhealth/SKILL.md
Commit: 36d8f13

Reviewed: about 6 hours ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.