reasonhub-terminology-crossmap

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Terminology Crossmap → SNOMED

术语交叉映射 → SNOMED

Overview

概述

ICD-10-CM, LOINC, and RxNorm all have strong use cases but limited semantic depth. SNOMED CT's attribute model (finding site, causative agent, associated morphology, etc.) is unmatched for answering clinical relationship questions.

This skill bridges the gap: given a code in any system, find its SNOMED equivalent and then use the

snomed-semantic

skill to query relationships.

No formal crossmap table is loaded. Mapping uses semantic search (
search_snomed
) on the source concept's display name. Always verify the match before proceeding.

ICD-10-CM、LOINC和RxNorm各有适用场景，但语义深度有限。SNOMED CT的属性模型（发病部位、致病因子、相关形态等）在解答临床关联问题方面表现卓越。

本技能可填补这一空白：给定任意系统的代码，找到对应的SNOMED等效代码，然后使用

snomed-semantic

技能查询关联关系。

未加载正式交叉映射表。映射通过对源概念显示名称执行语义搜索（
search_snomed
）实现。在继续操作前，请务必验证匹配结果。

When to Propose a Crossmap

何时建议交叉映射

Proactively suggest crossmapping to SNOMED when:

The user has an ICD-10-CM code and asks "what is related to this?"
The user has a LOINC observation code and wants to understand the clinical domain semantically
The user has an RxNorm ingredient and wants to find all disorders it treats or all procedures that use it
The user asks about "symptoms of X", "conditions caused by X", or "procedures for X" using a non-SNOMED code

在以下场景下，应主动建议映射到SNOMED：

用户拥有ICD-10-CM代码并询问“与此相关的内容有哪些？”
用户拥有LOINC观察代码，想要从语义层面理解临床领域
用户拥有RxNorm成分代码，想要查找其可治疗的所有病症或所有使用该成分的操作
用户使用非SNOMED代码询问“X的症状”“X引发的病症”或“针对X的操作”

Workflow

工作流程

Phase 1 — Look up the source code

阶段1 — 查询源代码

codesystem_lookup(code="<source_code>", system="<source_system>")

Extract the

display

. This is the search query for Phase 2.

codesystem_lookup("I25.10", "http://hl7.org/fhir/sid/icd-10-cm")

codesystem_lookup(code="<source_code>", system="<source_system>")

提取

display

字段，作为阶段2的搜索查询词。

codesystem_lookup("I25.10", "http://hl7.org/fhir/sid/icd-10-cm")

display: "Atherosclerotic heart disease of native coronary artery

without angina pectoris"

undefined

undefined

Phase 2 — Search SNOMED for the equivalent

阶段2 — 搜索SNOMED等效代码

Run this in parallel with Phase 1 if the source display is already known.

search_snomed(query="coronary arteriosclerosis disorder", top_k=5)

Selection heuristics:

ICD-10 diagnosis → prefer
```
(disorder)
```
semantic tag
LOINC observation → prefer
```
(observable entity)
```
RxNorm ingredient → prefer
```
(substance)
```
Prefer
```
sufficientlyDefined = true
```
(richer attributes)
When ambiguous, present 2–3 candidates and ask the user

如果已知源显示名称，可与阶段1并行执行此步骤。

search_snomed(query="coronary arteriosclerosis disorder", top_k=5)

选择规则：

ICD-10诊断代码 → 优先选择带
```
(disorder)
```
语义标签的概念
LOINC观察代码 → 优先选择带
```
(observable entity)
```
语义标签的概念
RxNorm成分代码 → 优先选择带
```
(substance)
```
语义标签的概念
优先选择
```
sufficientlyDefined = true
```
的概念（属性更丰富）
若存在歧义，展示2-3个候选选项并请用户选择

Phase 3 — Confirm the match and extract the pivot

阶段3 — 确认匹配并提取枢纽概念

codesystem_lookup(code="53741008", system="http://snomed.info/sct")

Check

inactive = false

and

display

matches. Then read the finding site attribute (

363698007

) directly from the response — this is the body structure concept ID you will use as the procedure site filter value in Phase 4.

undefined

codesystem_lookup(code="53741008", system="http://snomed.info/sct")

检查

inactive = false

且

display

匹配。然后直接从响应中读取发病部位属性（

363698007

）——这将作为阶段4中操作部位过滤值的身体结构概念ID。

undefined

Coronary arteriosclerosis (53741008)

363698007 (Finding site) = 41801008 (Coronary artery structure) ← pivot

363698007 (Finding site) = 41801008 (Coronary artery structure) ← 枢纽概念


> **This pivot step is the bridge between the diagnosis code and the
> procedure ValueSet.** The disorder’s finding site becomes the procedure’s
> site filter value.


> **此枢纽步骤是诊断代码与操作ValueSet之间的桥梁**。病症的发病部位将成为操作的部位过滤值。

Phase 4 — Discover the procedure site attribute

阶段4 — 确认操作部位属性

Before building the filter, look up one known representative procedure from the target domain to confirm which attribute it uses. Run this in parallel with Phase 3 if you already have a candidate procedure in mind.

codesystem_lookup("415070008", "http://snomed.info/sct")  # PCI

在构建过滤器之前，先查询目标领域中一个已知的代表性操作，确认其使用的属性。如果已确定候选操作，可与阶段3并行执行此步骤。

codesystem_lookup("415070008", "http://snomed.info/sct")  # PCI

363704007 (Procedure site - Direct) = 41801008 ← present

363704007 (Procedure site - Direct) = 41801008 ← 存在

405813007 (Procedure site - Indirect) = 41801008 ← also present

405813007 (Procedure site - Indirect) = 41801008 ← 也存在

codesystem_lookup("232717009", "http://snomed.info/sct") # CABG

363704007 (Procedure site - Direct) = — not present

363704007 (Procedure site - Direct) = — 不存在

405813007 (Procedure site - Indirect) = 41801008 ← only Indirect

405813007 (Procedure site - Indirect) = 41801008 ← 仅存在Indirect属性


**Rule:** use `405813007` (Indirect) when it appears on all representative
concepts. `363704007` (Direct) alone will miss procedures coded only to
Indirect. When unsure, look up two or three known procedures and check.


**规则：** 如果所有代表性概念都包含`405813007`（Indirect），则使用该属性。仅使用`363704007`（Direct）会遗漏仅编码Indirect属性的操作。若不确定，可查询2-3个已知操作进行验证。

Phase 5 — Build the ValueSet filter

阶段5 — 构建ValueSet过滤器

Use the confirmed site attribute + pivot concept ID + procedure hierarchy:

json

{
  "resourceType": "ValueSet",
  "name": "CoronaryArteryProcedures",
  "title": "Procedures on the Coronary Artery",
  "status": "draft",
  "compose": {
    "include": [{
      "system": "http://snomed.info/sct",
      "version": "<from list_available_codesystem_versions>",
      "filter": [
        { "property": "405813007", "op": "=",   "value": "41801008" },
        { "property": "concept",   "op": "is-a", "value": "71388002" },
        { "property": "inactive",  "op": "=",   "value": "false" }
      ]
    }]
  }
}

71388002
is the SNOMED root for Procedure. Always include it to avoid non-procedure concepts that may also encode a procedure site attribute.

使用确认的部位属性 + 枢纽概念ID + 操作层级：

json

{
  "resourceType": "ValueSet",
  "name": "CoronaryArteryProcedures",
  "title": "Procedures on the Coronary Artery",
  "status": "draft",
  "compose": {
    "include": [{
      "system": "http://snomed.info/sct",
      "version": "<from list_available_codesystem_versions>",
      "filter": [
        { "property": "405813007", "op": "=",   "value": "41801008" },
        { "property": "concept",   "op": "is-a", "value": "71388002" },
        { "property": "inactive",  "op": "=",   "value": "false" }
      ]
    }]
  }
}

71388002
是SNOMED中操作的根节点。务必包含该节点，以避免可能也编码操作部位属性的非操作概念。

Source System Guidance

源系统指南

ICD-10-CM → SNOMED

ICD-10-CM codes map well to SNOMED disorders and findings.

ICD-10 code type	Target SNOMED semantic tag
Diagnosis codes (A–Z chapters)	`(disorder)`
Symptom codes (R chapter)	`(finding)`
External cause codes (V–Y)	`(event)` or `(finding)`
Z codes (factors influencing health)	`(situation)` or `(finding)`

Tip: ICD-10 codes are intentionally coarser than SNOMED. A single ICD-10 code often maps to multiple SNOMED concepts. Choose the most clinically appropriate one for the query goal.

Example crossmap:

ICD-10:  E11.9  "Type 2 diabetes mellitus without complications"
SNOMED:  44054006  "Type 2 diabetes mellitus"  (disorder)

Once mapped, SNOMED lets you find:

All disorders with

causative agent = insulin resistance mechanism

All subtypes via
```
concept is-a 44054006
```
All finding sites affected:
```
363698007 = <pancreas / various>
```

ICD-10-CM代码可很好地映射到SNOMED的病症和发现类概念。

ICD-10代码类型	目标SNOMED语义标签
诊断代码（A-Z章节）	`(disorder)`
症状代码（R章节）	`(finding)`
外部原因代码（V-Y章节）	`(event)` 或 `(finding)`
Z代码（影响健康的因素）	`(situation)` 或 `(finding)`

提示： ICD-10代码的粒度故意比SNOMED更粗。一个ICD-10代码通常对应多个SNOMED概念。请根据查询目标选择最符合临床需求的概念。

映射示例：

ICD-10:  E11.9  "Type 2 diabetes mellitus without complications"
SNOMED:  44054006  "Type 2 diabetes mellitus"  (disorder)

完成映射后，可通过SNOMED查找：

所有

causative agent = insulin resistance mechanism

的病症

通过
```
concept is-a 44054006
```
查找所有子类型
所有受影响的发病部位：
```
363698007 = <pancreas / various>
```

RxNorm → SNOMED

RxNorm ingredients map to SNOMED substances or products.

RxNorm TTY	Target SNOMED semantic tag
`IN` (Ingredient)	`(substance)`
`BN` (Brand Name)	`(product)`
`SCD` (Semantic Clinical Drug)	`(product)`

Example crossmap:

RxNorm:  1049502  "metformin 500 MG Oral Tablet"
→ Strip to ingredient: metformin
SNOMED:  372567009  "Metformin"  (substance)

Once mapped, SNOMED lets you find:

Disorders where this substance is the causative agent (
```
246075003
```
)
Procedures that use this substance
Other substances in the same chemical class via hierarchy

RxNorm ingredient extraction tip: If the RxNorm code is a

SCD

SBD

(drug + dose form + strength), first extract just the ingredient name using

codesystem_lookup

then search SNOMED on the ingredient name alone.

RxNorm成分可映射到SNOMED的物质或产品类概念。

RxNorm TTY	目标SNOMED语义标签
`IN` （成分）	`(substance)`
`BN` （品牌名称）	`(product)`
`SCD` （语义临床药物）	`(product)`

映射示例：

RxNorm:  1049502  "metformin 500 MG Oral Tablet"
→ 提取成分：metformin
SNOMED:  372567009  "Metformin"  (substance)

完成映射后，可通过SNOMED查找：

以该物质为致病因子（
```
246075003
```
）的病症
使用该物质的操作
通过层级查找同一化学类别中的其他物质

RxNorm成分提取提示： 如果RxNorm代码是

SCD

SBD

（药物+剂型+剂量），请先使用

codesystem_lookup

提取成分名称，再仅以成分名称为关键词搜索SNOMED。

LOINC → SNOMED

LOINC observation codes map to SNOMED observable entities or procedures.

LOINC CLASS	Target SNOMED semantic tag
`CHEM` (Chemistry)	`(observable entity)` — but see caveat below
`HEM/BC` (Hematology)	`(observable entity)` — but see caveat below
`MICRO` (Microbiology)	`(procedure)`
`RAD` (Radiology)	`(procedure)` or `(observable entity)`

⚠️ SNOMED chemistry analyte caveat: Most common chemistry lab analytes (glucose, creatinine, ALT, sodium, calcium, etc.) do not have active
(observable entity)
concepts in SNOMED CT. The best active pivots are almost always
(finding)
with a “level – finding” display, or
(procedure)
with a “measurement” display. Accept the best active match from the first
search_snomed
call regardless of semantic tag. Do not run additional search rounds specifically hunting for
(observable entity)
— it rarely exists for common chemistry analytes and the search will not find what isn’t there.

Example crossmap:

LOINC:   2339-0  "Glucose [Mass/volume] in Blood"
SNOMED:  33747003  "Blood glucose measurement"  (procedure)
         or
         434912009  "Blood glucose concentration"  (observable entity)

LOINC观察代码可映射到SNOMED的可观察实体或操作类概念。

LOINC类别	目标SNOMED语义标签
`CHEM` （化学）	`(observable entity)` — 但请注意下方提示
`HEM/BC` （血液学）	`(observable entity)` — 但请注意下方提示
`MICRO` （微生物学）	`(procedure)`
`RAD` （放射学）	`(procedure)` 或 `(observable entity)`

⚠️ SNOMED化学分析物提示： 大多数常见的化学实验室分析物（葡萄糖、肌酐、谷丙转氨酶、钠、钙等）在SNOMED CT中没有活跃的
(observable entity)
概念。最合适的活跃枢纽概念几乎都是带“level – finding”显示的
(finding)
，或带“measurement”显示的
(procedure)
。接受首次
search_snomed
调用返回的最佳活跃匹配结果，无需考虑语义标签。不要为了寻找
(observable entity)
而额外执行搜索——常见化学分析物通常不存在这类概念，搜索也无法找到不存在的内容。

映射示例：

LOINC:   2339-0  "Glucose [Mass/volume] in Blood"
SNOMED:  33747003  "Blood glucose measurement"  (procedure)
         或
         434912009  "Blood glucose concentration"  (observable entity)

Getting the COMPONENT LP code for a LOINC observation

获取LOINC观察代码的COMPONENT LP代码

The

COMPONENT

LP code (e.g.,

LP14635-4

for Glucose) is returned directly by

codesystem_lookup

on any LOINC observation code. Read it from the response — do not guess it, do not search loinc.org:

codesystem_lookup("2345-7", "http://loinc.org")

任意LOINC观察代码的

COMPONENT

LP代码（例如葡萄糖对应的

LP14635-4

）可直接通过

codesystem_lookup

获取。请从响应中读取该代码——不要猜测，也不要搜索loinc.org：

codesystem_lookup("2345-7", "http://loinc.org")

COMPONENT = LP14635-4 (Glucose) ← use this in COMPONENT filter

COMPONENT = LP14635-4 (Glucose) ← 在COMPONENT过滤器中使用此代码

SYSTEM = LP7576-4 (Ser/Plas) ← use this in SYSTEM filter

SYSTEM = LP7576-4 (Ser/Plas) ← 在SYSTEM过滤器中使用此代码

undefined

undefined

Getting the members of a LOINC panel

获取LOINC面板的成员代码

codesystem_lookup

on a panel code returns the panel’s own axes but NOT its members. To get members, expand with a

panel-parent

filter:

json

{
  "resourceType": "ValueSet",
  "compose": {
    "include": [{
      "system": "http://loinc.org",
      "version": "<version>",
      "filter": [
        { "property": "panel-parent", "op": "=", "value": "24323-8" }
      ]
    }]
  }
}

Expand this with

reasonhub-skills expand

to get the full member list.

The result includes the panel code itself and any nested panel codes — filter to unique codes before processing. Then call

codesystem_lookup

on each member to get its

COMPONENT

and

SYSTEM

LP codes. Do this via the API — do not fetch individual loinc.org/CODE pages to extract LP codes.

codesystem_lookup("2345-7", "http://loinc.org")

对面板代码执行

codesystem_lookup

会返回面板自身的轴信息，但不会返回其成员代码。要获取成员代码，请使用

panel-parent

过滤器扩展：

json

{
  "resourceType": "ValueSet",
  "compose": {
    "include": [{
      "system": "http://loinc.org",
      "version": "<version>",
      "filter": [
        { "property": "panel-parent", "op": "=", "value": "24323-8" }
      ]
    }]
  }
}

使用

reasonhub-skills expand

扩展以获取完整成员列表。

结果包含面板代码本身及所有嵌套面板代码——处理前请过滤出唯一代码。然后对每个成员调用

codesystem_lookup

以获取其

COMPONENT

和

SYSTEM

LP代码。请通过API执行此操作——不要抓取loinc.org/CODE页面来提取LP代码。

codesystem_lookup("2345-7", "http://loinc.org")

COMPONENT = LP14635-4 ← use this in COMPONENT filter

COMPONENT = LP14635-4 ← 在COMPONENT过滤器中使用此代码

SYSTEM = LP7576-4 ← use this in SYSTEM filter

SYSTEM = LP7576-4 ← 在SYSTEM过滤器中使用此代码


To find sibling observations sharing the same `COMPONENT`, use
`reasonhub-skills expand` with a `COMPONENT = <LP_CODE>` filter —
**not `search_loinc`**. `search_loinc` returns semantically similar
observations, not structurally related ones sharing the same LOINC axis.

```json
{
  "resourceType": "ValueSet",
  "compose": {"include": [{
    "system": "http://loinc.org",
    "version": "<version>",
    "filter": [
      { "property": "COMPONENT", "op": "=", "value": "<LP_CODE_FROM_LOOKUP>" },
      { "property": "STATUS",    "op": "=", "value": "ACTIVE" }
    ]
  }]}
}

⛔ Do not scrape loinc.org pages to get panel structure or LP codes. Do not use
search_loinc
to find LP codes for analytes —
search_loinc
returns semantic proximity matches, not the exact LP code bound to a given LOINC observation. The LP code is in the
codesystem_lookup
response. Do not use
search_loinc
for sibling finding by COMPONENT or SYSTEM axis. All three are available through structured API calls.


要查找共享同一`COMPONENT`的同类观察代码，请使用`reasonhub-skills expand`并设置`COMPONENT = <LP_CODE>`过滤器——**不要使用`search_loinc`**。`search_loinc`返回语义相似的观察代码，而非共享同一LOINC轴的结构相关代码。

```json
{
  "resourceType": "ValueSet",
  "compose": {"include": [{
    "system": "http://loinc.org",
    "version": "<version>",
    "filter": [
      { "property": "COMPONENT", "op": "=", "value": "<LP_CODE_FROM_LOOKUP>" },
      { "property": "STATUS",    "op": "=", "value": "ACTIVE" }
    ]
  }]}
}

⛔ 请勿抓取loinc.org页面以获取面板结构或LP代码。请勿使用
search_loinc
查找分析物的LP代码——
search_loinc
返回语义相近的匹配结果，而非与给定LOINC观察代码绑定的精确LP代码。LP代码可在
codesystem_lookup
响应中获取。请勿使用
search_loinc
通过COMPONENT或SYSTEM轴查找同类代码。以上三种操作均可通过结构化API调用完成。

Handling Ambiguous Mappings

处理歧义映射

When

search_snomed

returns multiple plausible matches:

Narrow with a more specific query: add the semantic domain (
```
"myocardial infarction disorder"
```
,
```
"glucose substance"
```
)
Compare
```
parent
```
properties — the right SNOMED concept sits under the expected parent hierarchy

Use

codesystem_subsumes

to verify the candidate is in the right subtree:

codesystem_subsumes(
  code_a="64572001",   # Disease
  code_b="<candidate>",
  system="http://snomed.info/sct"
)

If still ambiguous, present candidates with their
```
display
```
,
```
semanticTag
```
, and
```
parent
```
names. Let the user choose.

当

search_snomed

返回多个合理匹配结果时：

使用更具体的查询词缩小范围：添加语义领域（例如
```
"myocardial infarction disorder"
```
、
```
"glucose substance"
```
）
对比
```
parent
```
属性——正确的SNOMED概念应位于预期的父层级下

使用

codesystem_subsumes

验证候选概念是否属于正确的子树：

codesystem_subsumes(
  code_a="64572001",   # Disease
  code_b="<candidate>",
  system="http://snomed.info/sct"
)

若仍存在歧义，展示候选概念的
```
display
```
、
```
semanticTag
```
和
```
parent
```
名称，让用户选择。

Full Example: I25.10 → SNOMED → Coronary Procedure ValueSet

完整示例：I25.10 → SNOMED → 冠脉操作ValueSet

User request: "We have I25.10 in our encounter data and need a ValueSet of coronary procedures — PCI, CABG, angiography, stent placement."

Phase 1 — Look up ICD-10 display:

codesystem_lookup("I25.10", "http://hl7.org/fhir/sid/icd-10-cm")

用户请求： "我们的就诊数据中有I25.10代码，需要一个包含冠脉操作的ValueSet——PCI、CABG、血管造影、支架置入。"

阶段1 — 查询ICD-10显示名称：

codesystem_lookup("I25.10", "http://hl7.org/fhir/sid/icd-10-cm")

display: "Atherosclerotic heart disease of native coronary artery

without angina pectoris"


**Phase 2 — Search SNOMED (run in parallel with Phase 1 if display is known):**

search_snomed("coronary arteriosclerosis disorder", top_k=5)


**阶段2 — 搜索SNOMED（若已知显示名称，可与阶段1并行执行）：**

search_snomed("coronary arteriosclerosis disorder", top_k=5)

Top match: 53741008 "Coronary arteriosclerosis" (disorder)

最佳匹配：53741008 "Coronary arteriosclerosis" (disorder)


**Phase 3 — Confirm and extract pivot (run in parallel with a known procedure lookup):**

codesystem_lookup("53741008", "http://snomed.info/sct")


**阶段3 — 确认匹配并提取枢纽概念（与已知操作查询并行执行）：**

codesystem_lookup("53741008", "http://snomed.info/sct")

sufficientlyDefined = true

363698007 (Finding site) = 41801008 (Coronary artery structure) ← pivot

363698007 (Finding site) = 41801008 (Coronary artery structure) ← 枢纽概念


**Phase 4 — Discover procedure site attribute:**


**阶段4 — 确认操作部位属性：**

Run in parallel with Phase 3

与阶段3并行执行

codesystem_lookup("415070008", "http://snomed.info/sct") # PCI

363704007 (Direct) = 41801008 ← present

363704007 (Direct) = 41801008 ← 存在

405813007 (Indirect) = 41801008 ← present

405813007 (Indirect) = 41801008 ← 存在

codesystem_lookup("232717009", "http://snomed.info/sct") # CABG

363704007 (Direct) = — not present

363704007 (Direct) = — 不存在

405813007 (Indirect) = 41801008 ← only Indirect

405813007 (Indirect) = 41801008 ← 仅存在Indirect属性

→ use 405813007 — it covers both PCI and CABG

→ 使用405813007 — 可同时覆盖PCI和CABG


**Phase 5 — ValueSet output:**
```json
{
  "resourceType": "ValueSet",
  "name": "CoronaryArteryProcedures",
  "title": "Procedures on the Coronary Artery (SNOMED CT)",
  "status": "draft",
  "compose": {
    "include": [{
      "system": "http://snomed.info/sct",
      "version": "http://snomed.info/sct/731000124108/version/20250901",
      "filter": [
        { "property": "405813007", "op": "=",   "value": "41801008" },
        { "property": "concept",   "op": "is-a", "value": "71388002" },
        { "property": "inactive",  "op": "=",   "value": "false" }
      ]
    }]
  }
}


**阶段5 — ValueSet输出：**
```json
{
  "resourceType": "ValueSet",
  "name": "CoronaryArteryProcedures",
  "title": "Procedures on the Coronary Artery (SNOMED CT)",
  "status": "draft",
  "compose": {
    "include": [{
      "system": "http://snomed.info/sct",
      "version": "http://snomed.info/sct/731000124108/version/20250901",
      "filter": [
        { "property": "405813007", "op": "=",   "value": "41801008" },
        { "property": "concept",   "op": "is-a", "value": "71388002" },
        { "property": "inactive",  "op": "=",   "value": "false" }
      ]
    }]
  }
}

After Crossmapping

交叉映射完成后

Once you have the SNOMED concept ID, follow the

snomed-semantic

skill for the full set of relationship query patterns.

获取SNOMED概念ID后，请遵循

snomed-semantic

技能的完整关联查询模式进行操作。

Output

输出内容

Every crossmap delivers two things.

每次交叉映射都会生成以下两类内容。

1. Crossmap provenance (always)

1. 交叉映射溯源（必选）

Show the mapping chain so the user can verify it:

Step	Code	System	Display
Source	`I25.10`	ICD-10-CM	Atherosclerotic heart disease of native coronary artery without angina pectoris
SNOMED match	`53741008`	SNOMED CT	Coronary arteriosclerosis
Pivot (finding site)	`41801008`	SNOMED CT	Coronary artery structure

展示映射链，方便用户验证：

步骤	代码	系统	显示名称
源	`I25.10`	ICD-10-CM	Atherosclerotic heart disease of native coronary artery without angina pectoris
SNOMED匹配结果	`53741008`	SNOMED CT	Coronary arteriosclerosis
枢纽概念（发病部位）	`41801008`	SNOMED CT	Coronary artery structure

2. FHIR ValueSet JSON + optional expansion

2. FHIR ValueSet JSON + 可选扩展

Deliver the complete

ValueSet

resource. Then ask:

"Would you like me to expand this to preview the matching procedure codes? I can show results as a markdown table or CSV."

Attempt

valueset_expand

once if the user says yes. If expansion succeeds, check the response for a

total

count. The MCP transport layer truncates returned rows regardless of the

count

parameter, and

offset

-based paging is unreliable. If rows returned are fewer than
total
, label the output and stop:

⚠️ Partial result — {n} of {total} codes shown. The full set is defined by the ValueSet JSON above.

Do not retry with different

count

offset

values.

CSV format:

csv

code,display
415070008,"Percutaneous coronary intervention"
232717009,"Coronary artery bypass graft"
33367005,"Angiography of coronary artery"

提供完整的

ValueSet

资源。然后询问：

"是否需要扩展此ValueSet以预览匹配的操作代码？我可以将结果以Markdown表格或CSV格式展示。"

如果用户同意，尝试执行一次

valueset_expand

。如果扩展成功，检查响应中的

total

计数。MCP传输层会忽略

count

参数并截断返回的行数，且基于

offset

的分页不可靠。如果返回的行数少于
total
，请标记输出并停止：

⚠️ 部分结果 — 显示{n}条，共{total}条代码。完整集合由上方的ValueSet JSON定义。

请勿尝试使用不同的

count

或

offset

值重试。

CSV格式示例：

csv

code,display
415070008,"Percutaneous coronary intervention"
232717009,"Coronary artery bypass graft"
33367005,"Angiography of coronary artery"