Back to Details

health-human-factors

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

EHR Design Review

EHR设计评审

Overview

概述

Use this skill to inspect healthcare or EHR software screens, components, mockups, or code and produce a structured report of design issues mapped to established healthcare usability and safety standards. The review covers patient identity, layout, color, typography, data display, numeric formatting, units, dates, alerts, medication safety, forms, accessibility, workflow, audit logging, error prevention, terminology, interoperability, internationalization, security, and documentation.
使用本技能检查医疗保健或EHR软件界面、组件、原型或代码，生成结构化的设计问题报告，问题将对应已确立的医疗保健可用性与安全标准。评审范围涵盖患者身份、布局、颜色、排版、数据展示、数值格式、单位、日期、警报、用药安全、表单、可访问性、工作流、审计日志、错误预防、术语、互操作性、国际化、安全以及文档。

Operating Rules

运行规则

  • Never change code, designs, configurations, or documentation.
  • Do not present the output as a formal certification or regulatory determination.
  • Bias toward observable evidence from the artifacts under review and clearly separate:
    • confirmed violations from the code, markup, design, or config
    • likely inferences from surrounding implementation
    • areas that require runtime testing, user research, or policy validation
  • When a guideline cannot be evaluated from the provided artifacts, mark it as not assessable rather than passing or failing.
  • 永远不要修改代码、设计、配置或文档。
  • 不要将输出内容作为正式认证或监管判定结果。
  • 优先基于被评审工件中的可观察证据，明确区分：
    • 从代码、标记、设计或配置中确认的违规项
    • 从周边实现推断的可能问题
    • 需要运行时测试、用户研究或政策验证的领域
  • 当某条指南无法基于提供的工件进行评估时，将其标记为不可评估，而非判定为通过或不通过。

Workflow

工作流

  1. Confirm the scope: which screens, components, modules, or code paths to review.
  2. Load 
    references/style-guide.md
    to access the full design criteria.
  3. Walk through each review category (see categories below) against the artifacts in scope.
  4. Assign severity and confidence for each finding.
  5. Produce a report only. Do not draft fixes, patches, or redesigns unless explicitly asked.
  1. 确认评审范围：需要评审的界面、组件、模块或代码路径。
  2. 加载
    references/style-guide.md
    获取完整设计准则。
  3. 对照范围内的工件，逐一检查每个评审类别（见下方类别列表）。
  4. 为每个发现的问题分配严重等级和置信度。
  5. 仅生成报告。除非明确要求，否则不要起草修复方案、补丁或重设计内容。

Review Categories

评审类别

Each category maps to a section of the style guide reference.
  1. Patient Context and Identity — persistent header, required identifiers, patient-switch confirmation, environment indicators
  2. Layout and Information Hierarchy — summary order, navigation depth, click efficiency, cross-module consistency
  3. Color Standards — semantic color use, dual-coding (never color-only), WCAG contrast ratios
  4. Typography — font legibility, size minimums, avoidance of condensed or decorative fonts
  5. Data Tables and Clinical Data Display — alignment, column consistency, sorting, filtering, reference ranges, abnormal value marking
  6. Numeric Formatting — thousands separators, decimal precision, trailing zeros
  7. Units of Measure — units always displayed, UCUM preference, no bare numbers
  8. Date and Time Formatting — ISO-8601 storage, unambiguous display (DD Mon YYYY), 24-hour time
  9. Alerts and Clinical Decision Support — alert levels, fatigue prevention, override documentation, clear explanations
  10. Medication Safety — dangerous abbreviation avoidance, structured order display, trailing-zero prevention
  11. Forms and Data Entry — structured input preference, autocomplete, range display, immediate validation
  12. Accessibility — WCAG 2.1 AA compliance, keyboard navigation, screen reader support, focus indicators, no hover-only information
  13. Workflow Optimization — click reduction, persistent key data, minimal modals, quick patient navigation
  14. Audit Logging — user, timestamp, action, before/after data, location in audit records
  15. Error Prevention — proactive constraints, range warnings, input validation before submission
  16. Clinical Terminology Standards — SNOMED CT, LOINC, ICD-10 usage for coded concepts
  17. Interoperability and Data Exchange — HL7 FHIR resource alignment, structured data exchange
  18. Internationalization — multi-language support, locale-aware formatting, standardized internal representation
  19. Security and Privacy — RBAC, session timeouts, encryption, audit logs, HIPAA alignment
  20. Documentation and Help — contextual help, error explanations, training materials, workflow guides
每个类别对应风格指南参考文档的一个章节。
  1. 患者上下文与身份 — 固定头部、必填标识、患者切换确认、环境指示器
  2. 布局与信息层级 — 摘要排序、导航深度、点击效率、跨模块一致性
  3. 颜色标准 — 语义化颜色使用、双重编码（禁止仅用颜色标识）、WCAG对比度要求
  4. 排版 — 字体易读性、最小字号、避免使用压缩或装饰性字体
  5. 数据表与临床数据展示 — 对齐方式、列一致性、排序、筛选、参考范围、异常值标记
  6. 数值格式化 — 千位分隔符、小数精度、尾随零处理
  7. 度量单位 — 始终展示单位、优先使用UCUM、禁止无单位裸数值
  8. 日期与时间格式化 — ISO-8601存储、无歧义展示（DD Mon YYYY）、24小时制
  9. 警报与临床决策支持 — 警报等级、疲劳预防、覆写记录、清晰说明
  10. 用药安全 — 避免危险缩写、结构化医嘱展示、防止尾随零风险
  11. 表单与数据录入 — 优先使用结构化输入、自动补全、范围展示、即时校验
  12. 可访问性 — WCAG 2.1 AA合规、键盘导航、屏幕阅读器支持、焦点指示器、禁止仅悬停展示信息
  13. 工作流优化 — 减少点击次数、固定关键数据、最小化弹窗、快速患者导航
  14. 审计日志 — 审计记录中的用户、时间戳、操作、变更前后数据、位置信息
  15. 错误预防 — 前置约束、范围预警、提交前输入校验
  16. 临床术语标准 — 编码概念使用SNOMED CT、LOINC、ICD-10
  17. 互操作性与数据交换 — 符合HL7 FHIR资源规范、结构化数据交换
  18. 国际化 — 多语言支持、本地化感知格式化、标准化内部表示
  19. 安全与隐私 — RBAC、会话超时、加密、审计日志、符合HIPAA要求
  20. 文档与帮助 — 上下文帮助、错误说明、培训材料、工作流指南

Constraints

约束

  • Review only — do not propose code changes unless the user explicitly requests remediation guidance.
  • Stay within the scope described in the frontmatter.
  • Surface patient-safety implications with the highest priority.
  • Distinguish between must-fix safety issues and nice-to-have improvements.
  • When artifacts are insufficient to evaluate a category, say so rather than guessing.
  • 仅做评审 — 除非用户明确要求修复指导，否则不要提议代码修改。
  • 严格遵循前置说明中定义的评审范围。
  • 最高优先级披露患者安全相关影响。
  • 区分必须修复的安全问题和锦上添花的优化项。
  • 当工件不足以评估某个类别时，明确说明而非猜测。

Resources

资源

  • references/style-guide.md
    : full Healthcare Software Design Style Guide with criteria, examples, and source standards
  • examples/example-report.md
    : example review report showing expected output shape, finding format, and coverage matrix
  • references/style-guide.md
    : 完整的医疗软件设计风格指南，包含评审准则、示例和参考标准来源
  • examples/example-report.md
    : 评审报告示例，展示预期输出结构、问题格式和覆盖矩阵

Invocation Modes

调用模式

Standalone (default)

独立模式（默认）

When invoked directly by a user or without the phrase "scoped review," operate normally: confirm scope interactively, load references, walk review categories, and produce the full report described in the Output Contract below.
当由用户直接调用或调用时未带「scoped review」短语时，按正常模式运行：交互确认评审范围、加载参考文档、遍历评审类别、生成下文输出约定中描述的完整报告。

Scoped

范围模式

When invoked with the phrase "scoped review" and a pre-determined list of file paths, operate in scoped mode:
  • Input: a list of file paths to review. Scope is pre-determined — do not ask for confirmation.
  • Behavior: skip interactive scope confirmation. Skip executive summary and coverage matrix generation. Review only the provided files against the review categories.
  • Output: return a findings-only list. Each finding uses this format:
    ### [HF-{n}] {title}
- Severity: critical | major | minor | info
- Category: {category from the 20 review categories}
- File: {path}:{line}
- Detail: {what was observed}
- Guideline: {which standard or rule applies}
    If no findings are discovered, return a single line: "No human-factors findings for the provided files."
当调用时带「scoped review」短语和预定义的文件路径列表时，按范围模式运行：
  • 输入：待评审的文件路径列表。范围已预先确定 — 无需询问确认。
  • 行为：跳过交互范围确认步骤，跳过执行摘要和覆盖矩阵生成，仅针对提供的文件对照评审类别进行检查。
  • 输出：仅返回问题列表。每个问题遵循以下格式：
    ### [HF-{n}] {title}
- Severity: critical | major | minor | info
- Category: {category from the 20 review categories}
- File: {path}:{line}
- Detail: {what was observed}
- Guideline: {which standard or rule applies}
    如果未发现任何问题，返回单行内容：「No human-factors findings for the provided files.」

Output Contract

输出约定

When operating in standalone mode, return a review report with:
  • Executive Summary: overall assessment, highest-risk findings, and scope of review
  • Scope: artifacts reviewed, categories assessed, categories not assessable
  • Findings Table with columns:
    • ID
    • Severity (critical / major / minor / info)
    • Category (from the 20 review categories)
    • Location (file, screen, component, or line reference)
    • Finding (what was observed)
    • Guideline (which standard or rule applies)
    • Risk (patient-safety or usability impact)
    • Confidence (confirmed / likely / unclear)
  • Category Coverage Matrix: for each of the 20 categories — compliant, partial, non-compliant, or not assessable
  • Positive Observations: areas where the design follows the guidelines well
  • Open Questions: areas requiring runtime testing, user research, or additional artifacts
  • Standards Basis: list of standards referenced in the review
在独立模式下运行时，返回的评审报告包含：
  • 执行摘要：整体评估、最高风险问题、评审范围
  • 范围说明：评审的工件、已评估类别、不可评估类别
  • 问题表包含以下列：
    • ID
    • 严重等级（critical / major / minor / info）
    • 类别（来自20个评审类别）
    • 位置（文件、界面、组件或行号参考）
    • 问题描述（观察到的内容）
    • 参考准则（适用的标准或规则）
    • 风险（患者安全或可用性影响）
    • 置信度（confirmed / likely / unclear）
  • 类别覆盖矩阵：针对20个类别分别标注 — 合规、部分合规、不合规、不可评估
  • 正面观察：设计符合指南要求的优秀部分
  • 待确认问题：需要运行时测试、用户研究或补充工件的领域
  • 标准依据：本次评审引用的标准列表