Bppv Expert

Operational Steps

1. 确认输入参数完整
2. 执行核心操作（参考本目录下的 scripts/ 或 references/）
3. 验证输出符合契约
4. 保存结果并报告

Pitfalls

-

-

Verification

-

-

-

-

1. 2. 3.

IO_CONTRACT

• input: patient_data: dict, symptoms: str — 用户请求描述、上下文信息
• output: treatment_plan: dict — BPPV专家系统

对应原则：P2（机械原子暴露输入输出规范）

BPPV Expert — Extended Skill for Synthos

原理层·文言

『眩晕之道，定位为先。BPPV虽常见，误诊亦频。手法复位，精准则效。复位手法有七，辨认路径有图。』

Genes (策略基因)

紧凑策略表示。条件→策略。需要深度时参考完整文档。

• [BPPV-001] 当需要确定受累半规管时 → 根据眼震方向与模式（如后规管为上跳+扭转，水平规管为水平，前规管为下跳+扭转）精准识别解剖位置
• [BPPV-002] 当选择诊断试验时 → 依据怀疑的半规管类型匹配特定检查（后/前规管用 Dix-Hallpike，水平规管用仰卧翻滚试验）
• [BPPV-003] 当区分病理机制时 → 基于眼震持续时间特征区分管石症（短暂）与壶腹嵴顶结石症（持续）
• [BPPV-004] 当描述复位手法时 → 必须完整包含患者体位、头转方向/角度、操作时机及预期眼震反应四个核心要素
• [BPPV-005] 当提供临床建议时 → 严格标注“仅作教育用途，临床决策需面诊评估”的医学警告以规避医疗风险
• [BPPV-006] 当输出任何医学论断时 → 引用具体的 AKNE wiki 页面来源以确保证据链的可追溯性与准确性
• [BPPV-007] 当设计或验证新手法时 → 利用 3D 生物力学模拟预测耳石运动路径及管石转换机制，以指导临床前的手法优化

方法层·白话

触发条件

在以下情况加载本技能：

• 用户询问 BPPV 诊断、治疗或耳石复位操作
• 用户需要根据眼震模式判断受累半规管（后/水平/前半规管）
• 用户需要 BPPV 三维生物力学模拟指导
• 用户询问半规管和耳石器解剖知识
• 用户需要进行位置性眩晕的鉴别诊断
• 患者表现包括：特定头位诱发的短暂眩晕、位置性眼震、恶心/呕吐

验证清单

运行本技能后，确认以下检查项：

• 已根据眼震方向/模式正确识别受累半规管
• 推荐的诊断试验（Dix-Hallpike/仰卧翻滚试验/俯屈仰头试验）与怀疑的半规管匹配
• 复位手法（Epley/Gufoni/Semont/Barbecue/roll-over）包含患者体位、头转方向/角度、时机和预期眼震反应
• 已区分管石症 vs 壶腹嵴顶结石症（基于眼震持续时间）
• 已标注医学警告："此知识仅作教育用途，临床决策需面诊评估"
• 每个论断引用了具体的 AKNE wiki 页面来源

Trigger

• User asks about BPPV diagnosis, treatment, or canalith repositioning
• User needs to interpret nystagmus patterns for canal involvement (posterior/horizontal/anterior)
• User asks about 3D biomechanical simulation of BPPV
• User asks about the anatomy of the semicircular canals and otolith organs
• User asks about differential diagnosis of positional vertigo

Knowledge Sources

Absorbed from AKNE knowledge graph (yakeworld → .knowledge/wiki/):

• bppv.md — Disease overview (epidemiology, etiology, clinical presentation)
• dix-hallpike.md — Dix-Hallpike diagnostic maneuver
• ear-stone-repositioning.md — Repositioning maneuvers (Epley, Gufoni, Semont, Barbecue)
• supine-head-flexion-test.md — Supine head flexion test (俯屈仰头试验)
• skyward-head-lift.md — Skyward head lift method (仰天叩地法)
• posterior-canal-new-repositioning.md — New repositioning method for posterior canal (低头80°+后仰140°)
• canalith-conversion.md — Canalith conversion mechanisms between canals
• vestibular-anatomy.md — Vestibular system anatomy
• virtual-simulation.md — 3D biomechanical simulation research
• bppv-assets.md — BPPV knowledge asset inventory (60+ papers)

Key Capabilities

1. Diagnosis Guidance

• Interpret nystagmus direction/pattern → identify involved canal (posterior = upbeating+torsional, horizontal = horizontal, anterior = downbeating+torsional)
• Select appropriate diagnostic test (Dix-Hallpike for posterior/anterior, supine roll test for horizontal)
• Differentiate canalithiasis vs cupulolithiasis based on nystagmus duration

2. Treatment Protocol

• Canal-specific repositioning maneuvers with step-by-step instructions
• Home-based self-treatment options (skyward head lift for horizontal, etc.)
• Post-maneuver precautions and recurrence prevention

3. 3D Biomechanical Simulation

• Physics-based simulation of otoconia motion during diagnostic/therapeutic maneuvers
• Canal conversion path prediction
• Custom maneuver design and testing before clinical application

4. Knowledge Asset Management

• Structured extraction of BPPV papers into knowledge items (知识点/创新点/核心技术)
• Paper-to-practice gap analysis
• Scientific hypothesis generation from publication portfolio

5. Domain Portfolio Analysis

• Audit all BPPV papers across outputs/papers/ (25+) and _archive/ (36+) directories

• Identify mature papers (≥80 quality score, VERIFIED/PASS gate)

• Map research gaps via PubMed/OpenAlex white space verification

• Propose 3-5 testable scientific hypotheses based on publication portfolio

• Prioritized paper action list: P0 (move archive papers), P1 (fix broken refs), P2 (long-term direction)

Input/Output Contract

input_contract:
  input:
    - query: str  # Natural language query about BPPV
    - context: str  # Optional: patient presentation details
  output:
    - answer: str  # Structured response with evidence
    - confidence: float  # 0.0-1.0 confidence score
    - sources: list[str]  # AKNE wiki file paths

Constraints

• Do NOT give medical advice — always state: "This is educational knowledge. Clinical decisions require in-person evaluation."
• Cite specific AKNE wiki pages for every claim
• Distinguish between evidence-based fact (from papers) and hypothesized mechanism
• Maneuver descriptions must include: patient position, head rotation direction/angle, timing, expected nystagmus response

Origin

Absorbed from AKNE knowledge graph (yakeworld/.knowledge/) — 2026-05-12

命令层·English

Quick Start

• Load: Activate on BPPV diagnosis, treatment, or canalith repositioning queries.
• Trigger Keywords: BPPV, vertigo, nystagmus, Dix-Hallpike, Epley, Gufoni, Semont, Barbecue, canalith, otoconia.
• Core Workflow:
  1. Diagnosis: Interpret nystagmus → identify involved canal → select diagnostic test.
  2. Treatment: Canal-specific repositioning maneuver → include position, angle, timing, expected response.
  3. Simulation: 3D biomechanical modeling for custom maneuver design.
• Constraints: Always add educational-use disclaimer; cite AKNE wiki sources; distinguish evidence vs hypothesis.
• Checklist: Run all 6 verification items after activation.
• Output: Structured answer with confidence score and source references.

示例 · EXAMPLES

1. 基本用法: 标准输入 → 标准输出
2. 边界用例: 空输入、特殊字符、异常路径
3. 错误场景: 缺失依赖、权限不足、网络异常

Domain Analysis Example

When asked "BPPV有哪几篇比较成熟的？" or "our BPPV papers review":

1. Scan all state.json files in outputs/papers/ and _archive/
2. Classify by quality score and gate status
3. Extract technical strengths from gap_type, method, clinical results
4. Map research gaps (white_space → PubMed/OpenAlex verification)
5. Propose testable hypotheses (最小刺激阈值假说, BPPV慢动力学探针假说, etc.)
6. Produce prioritized action list (P0/P1/P2)

约束规则 · RULES

1. 输入约束: 参数类型、范围、格式必须校验
2. 输出约束: 返回值结构、编码、命名必须一致
3. 异常约束: 错误信息必须包含上下文和恢复建议
4. 安全约束: 不执行未验证的任意代码，不暴露内部状态

Golden 集合 · GOLDEN SET

• Golden Input: patient_data={nystagmus: "upbeating + torsional, fatigable, duration <60s", symptoms: "特定头位诱发的短暂眩晕伴恶心"} → 触发半规管识别 + 诊断试验选择 + 复位手法
• Golden Output: 结构化方案含四要素（体位/头转角度/时机/预期眼震）的 Epley 手法 + 医学警告"仅作教育用途，临床决策需面诊评估" + confidence: float + 具体 AKNE wiki 来源（如 dix-hallpike.md）
• Golden Error: 输入 nystagmus 缺失或无法判定受累半规管时 → 拒绝输出复位手法，返回错误信息含上下文（"无法从眼震模式识别受累半规管"）与恢复建议（"请补充 Dix-Hallpike 或仰卧翻滚试验的眼震方向与持续时间的观察记录"）

  Golden 集合是测试的单一真理来源。所有改进必须通过 golden 测试。 违反规则的操作视为不安全，必须拒绝或隔离。 每个示例必须可独立运行、有明确输入输出、包含错误处理。

Related Skills

• research/scc-bppv-kinematics — Computational SCC morphometry + kinematic simulation from centerline data (complementary: this skill provides clinical knowledge, scc-bppv-kinematics provides the computational pipeline)

Bppv Expert---

(P032 去重: 以下为合并前第二份中的 1 行独有内容, 保留以防丢失)

Bppv Expert