radiology-translation
Design clinical use, reader/impact studies, human factors, deployment and lifecycle evidence.
Install / Use
npx skills add huang-sir1/radiology-skills --skill radiology-translationInstalls into whichever agent you are using.
SKILL.md
Installable skill definition
Quality Score
Category
OperationsSupported Platforms
Our assessment of radiology-translation
radiology-translation scores 84/100 on our quality scale, 567th of 746 Operations skills we index.
Its SKILL.md is 13 KB long, well organised into 10 sections and no code examples: a thorough specification that gives an agent plenty to work with.
With 1,785 GitHub stars, it is one of the more widely adopted skills in the catalogue.
Maintenance, license and trust
- The repository was last updated about 2 months ago, so radiology-translation is actively maintained.
- Our last check on 2026-11-05 found the source still online.
- No license is declared. By default that means all rights are reserved: you can read it, but reusing or redistributing it is not clearly permitted. Ask the author before building on it commercially.
- Its trust signals score 88/100, with 1 caution from licensing, adoption, age or documentation. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.
Safety scan
No issues foundOur scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful.
AI review by kimi-k2.7-code on 2026-10-01. Automated pattern scan on 2026-10-01. It catches known dangerous patterns, not every risk — read a skill before letting an agent act on it.
radiology-translation compared with similar skills
All 4 of these similar skills score higher than radiology-translation; compare them before choosing.
| Skill | Score | Stars | Updated | Format |
|---|---|---|---|---|
| radiology-translation (this skill)by huang-sir1 | 84 | 1.8k | 56d ago | SKILL.md |
| taste-skillby Leonxlnx | 100 | 116.8k | 8d ago | CLAUDE.md |
| Claude-Code-Game-Studiosby Donchitos | 100 | 39.7k | 7d ago | CLAUDE.md |
| x402by browser-use | 100 | 121.0k | 3d ago | SKILL.md |
| algorithmic-artby anthropics | 100 | 177.9k | 55d ago | SKILL.md |
Frequently asked questions
- How do I install radiology-translation?
- Run
npx skills add huang-sir1/radiology-skills --skill radiology-translation. The install tabs above show the steps for each supported agent. - Which AI agents does radiology-translation work with?
- It is written for Universal, as a SKILL.md file. Other agents that read the same format can often use it too.
- Is radiology-translation safe to use?
- Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful. It declares no license and scores 88/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
- Is radiology-translation still maintained?
- The repository was last updated about 2 months ago, so radiology-translation is actively maintained.
Skill content
View source on GitHubname: radiology-translation description: "Design clinical use, reader/impact studies, human factors, deployment and lifecycle evidence."
Clinical Translation, Reader Studies & Prospective Validation
Use this skill to determine which evidence is needed to move from retrospective performance toward a bounded claim about clinical use. The required design depends on the intended use and estimand: it may involve threshold/action analysis, a controlled reader study, silent prospective validation, active impact evaluation, implementation research or production monitoring.
Name note: this skill is clinical translation (bench-to-bedside evidence), not language translation. For Chinese→English manuscript translation/reconstruction →
radiology-writing(Chinese-author workflow).
Core stance
- Retrospective AUC ≠ clinical utility. Discrimination on a curated set says nothing about patient benefit. The claim must be earned with utility evidence, not asserted.
- Define the use scenario first. Screening, triage, diagnosis, staging, prognosis, treatment-response, surveillance, or MDT support — each fixes the output, the threshold, and the cost of errors.
- Reader studies are designed for the estimand. Choose crossed/partially crossed MRMC, sequential, parallel or another justified design; manage order, learning, memory and carryover, report reader/case sampling and match inference to the assignment structure (→ radiology-stats).
- Thresholds map to actions. Every operating point implies an action and consequence; decision-curve analysis models net benefit under explicit threshold/weighting assumptions but does not observe patient benefit or cost-effectiveness by itself.
- Match design to claim. Temporal, external, controlled-reader, silent prospective, pragmatic, quasi-experimental, randomized and monitoring designs answer different questions; there is no universal ranking that overrides estimand, bias, setting and feasibility.
- Translation is multi-axis. Technical performance, clinical validation, clinical impact, regulatory state, economic value, implementation and production monitoring remain separate. Never promote one because another axis is strong.
- Integrity & safety. Plan only; never claim deployment readiness without the evidence, and never give clinical or diagnostic recommendations for individual patients.
When to use
- "Design a reader study / clinician+AI gain study." / "帮我设计读者研究 / 医生+AI 增益研究。"
- "What does it take to claim clinical utility / move toward deployment?"
- "Plan a prospective / real-world validation." / "前瞻性验证、真实世界验证怎么设计?"
- "Map my model's threshold to a clinical action / net benefit."
- "Where does this sit in the PACS/RIS workflow?"
- "Is this site ready for silent deployment, local acceptance, active rollout, monitoring or retirement?" / "Which implementation-science framework and outcomes fit this study?"
Operating modes
Choose the narrowest mode: use-scenario, reader-study, clinical-impact,
regulatory-readiness, implementation-readiness, silent-deployment,
site-acceptance, implementation-evaluation, post-deployment-monitoring,
change-control/revalidation, or de-implementation/retirement.
When to open extra files
| File | Open when | |---|---| | references/use-scenario.md | Defining the clinical-use scenario, the model output, and the cost of errors | | references/reader-study.md | MRMC reader-study design: readers, washout, randomisation, with/without AI, outcomes | | references/threshold-to-action.md | Mapping operating points to actions; net-benefit / decision-curve; false-positive/negative consequences | | references/prospective-deployment.md | Prospective/real-world validation design; PACS/RIS integration; monitoring and drift | | references/regulatory-and-deployment-readiness.md | The user asks about clinical translation, FDA/EU/regulatory readiness, silent deployment, locked/adaptive models, model cards, lifecycle monitoring, or whether a retrospective AI/radiogenomics model can claim clinical use | | references/clinical-impact-study-design.md | Selecting controlled MRMC, sequential reading, silent prospective, pragmatic/cluster randomized, stepped-wedge, interrupted-time-series or active-deployment/RWE designs for the intended claim | | references/human-factors-and-implementation.md | Workflow fit, automation bias, alert burden, usability, training, overrides/abstention, implementation outcomes and sociotechnical failure analysis | | references/evidence-state-and-translation-lifecycle.md | Always when judging maturity/readiness or coordinating T0-T6; separates seven evidence axes and their claim ceilings | | references/implementation-science-and-scale-up.md | Selecting CFIR/NASSS, RE-AIM/PRISM, Proctor outcomes, ERIC strategies, hybrid designs, scale-up or de-implementation | | references/deployment-interoperability-and-site-acceptance.md | DICOM/IHE/PACS/RIS/EHR/worklist integration, output display/identity, failure fallback or end-to-end site acceptance | | references/production-monitoring-and-change-control.md | Active deployment monitoring, label-delay blind intervals, incidents/CAPA, change/revalidation, rollback, stop or retirement | | references/source-registry.md | Any rule-bearing regulatory, interoperability, human-factors, reporting or implementation claim; refresh volatile sources for the target context | | templates/clinical-translation-state-card.md | A durable seven-axis translation state and T0-T6 handoff artifact is requested | | templates/production-monitoring-contract.md | A monitor/release/event/change contract is requested |
Workflow
- Accept or create the clinical-domain lock. When disease-specific context matters, receive
from
radiology-clinical-domainthe population/pathway, decision point, intended use, reference-standard, treatment-time, subgroup/access, harm and claim-ceiling fields. Return there if any clinical fact or changing standard is unresolved; do not manufacture a pathway inside a reader-study plan. - Freeze the seven-axis state (
evidence-state-and-translation-lifecycle.md) and the exact intervention release. Select one current mode and the earliest unsupported decision; do not use a single readiness score. - Define the use scenario (use-scenario.md) — where in the pathway, who acts on the output, what the output is (risk score, stratification, mask, heatmap, report aid), error costs.
- Set the threshold-to-action map (threshold-to-action.md) — operating point(s), the action each triggers, and net-benefit framing across plausible thresholds.
- Select the clinical-impact design (clinical-impact-study-design.md). Do not use a controlled MRMC experiment to claim real-world outcome benefit, or an uncontrolled before–after rollout to claim causality.
- Design the reader study (reader-study.md) when reader performance is the estimand — MRMC or justified sequential workflow, case sampling, information availability, order/carryover/learning, alone vs +AI, abstention/failures and task-matched outcomes.
- Map human factors and implementation (human-factors-and-implementation.md): user roles, workstation/context, training, workload, alert/override behavior, automation bias, failure recovery, implementation outcomes and equity/access effects.
- For implementation evaluation/scale, open
implementation-science-and-scale-up.md; select frameworks by job, link strategies to determinant evidence and keep implementation outcomes separate from clinical outcomes. - For regulatory/deployment framing, open
regulatory-and-deployment-readiness.mdand bound the claim by intended use, model lock/adaptation, human oversight, monitoring, and evidence level. - Plan prospective/real-world validation (prospective-deployment.md) — temporal/prospective/ RWE design, workflow position, monitoring for drift, failure handling.
- For a real local integration or active release, run
deployment-interoperability-and-site-acceptance.mdand/orproduction-monitoring-and-change-control.md; preserve identity/display/fallback/site-test, denominator/label-delay/CAPA/rollback/retirement evidence. - Bound the claim — state exactly what level of evidence supports what level of claim;
route statistics to
radiology-statsand reporting toradiology-reporting.
Output contract
Use scenario— pathway position, decision-maker, output type, error costs.Seven-axis translation state— technical, clinical-validation, impact, regulatory, economic, implementation and production states; current T0-T6 position, stale evidence and claim ceiling.Threshold-to-action map— operating point(s) → action; net-benefit/DCA framing.Reader-study design(if applicable) — readers, washout, randomisation, arms, outcomes, statistic (MRMC → radiology-stats).Regulatory/deployment readiness— intended use, locked/adaptive status, oversight, change-control/revalidation, monitoring, and exact items needing live verification.Prospective/real-world plan— design, workflow integration, monitoring, drift.Human-factors/implementation plan— users, context, workload, training, alert/override/ abstention behavior, failure recovery, implementation outcomes and sociotechnical risks.Implementation-science contract— framework-role map, determinant evidence, strategies, implementation/clinical outcomes, adaptation/equity and scale/de-implementation boundary.Site-acceptance and production contract— when applicable: release, DICOM/IHE/system tests, monitoring denominators/thresholds, blind intervals, incidents/CAPA, rollback/revalidation/retirement.Evidence-to-claim matrix— each supplied design/evidence axis, its supported claim, assumptions/biases and prohibited stronger inference; do not force unlike designs onto one ladder.Evidence transition register— current evidence, next required study/artifact, success criterion, owner, and claim unlocked; preserve failures and drift/incident plans.
Quality bar
A good translation plan names the clinical decision, chooses a design and analysis matched to the estimand, ties thresholds to actions and explicit assumptions, plans the next decision-bearing evidence, and never lets retrospective performance masquerade as clinical readiness.
Maintainer validation
After changing a translation route, scientific boundary, source or template, run
scripts/validate_translation_skill.ps1. It checks the independent package, modes, boundaries,
sources, local links and prohibited over-absolute hierarchy language; it is a static contract test,
not evidence that a study, site or regulatory route passed.
Handoffs
- Disease/pathway, reference-standard, treatment-time, population representativeness and current
clinical-standard facts ↔
radiology-clinical-domain; this skill returns the reader/prospective/ threshold/deployment design and does not overwrite the supplied clinical lock. - MRMC / net-benefit / decision-curve statistics →
radiology-stats. - F
Truncated for display — read the full file on GitHub.
Related Skills
taste-skill
116.8kAgent skills that give coding agents taste: art direction, reference images and a browser check instead of generic AI pages
Claude-Code-Game-Studios
39.7kTurn Claude Code into a full game dev studio — 49 AI agents, 72 workflow skills, and a complete coordination system mirroring real studio hierarchy.
x402
121.0kSet up Browser Use Cloud payments with x402 — pay per request from a crypto wallet (USDC on Base mainnet), no signup or API key.
algorithmic-art
177.9kCreating algorithmic art using p5.js with seeded randomness and interactive parameter exploration. Use this when users request creating art using code, generative art, algorithmic art, flow fields, or particle systems.
Languages
Trust signals
From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.
