AI-处理健康卡 n8n

AI-处理健康卡 是一个 PDF4me 节点操作 n8n 读取健康保险卡并以结构化形式返回其详细信息 JSON. 使用它可以从扫描或照片中获取会员 ID、团体编号、计划类型和处方路由代码，而不是手动从卡片上输入。

此节点的功能

PDF4me： AI-处理健康卡 拍摄卡片图像或 PDF 并返回一个 健康数据 它读取的是对象而不是文件。它读取成员身份、计划详情以及 rxBin， rxPcn， 和 rxGroup 处方路由代码，并报告 每个字段的置信度得分 因此，不确定的读取结果可以转交给人工处理。它负责提取数据；至于如何存储和处理这些数据，则由您决定。

相关博客文章

目前尚无关于此功能的博客文章——敬请期待。

在此期间，您可以浏览 PDF4me 博客，查看适用于各平台的教程和工作流程。

访问博客 →

验证您的身份 API 要求

每一个 PDF4me 节点 n8n 需要有效 用于连接的凭证创建或选择一个可以容纳您的 PDF4me API 关键在于确保工作流程能够安全地验证提取请求。

Get your API Key

您不容错过的重要事实

提取信息并不等同于遵守规定。

此操作返回的值用于识别个人及其保险范围。将其导入您的系统后，这些个人将适用您现有的健康数据规则。此节点不会改变这一规则。

置信度得分是准确性信号。

响应中包含一个针对每个字段的置信度对象。这是该操作公开的唯一质量指标，因此应以此为依据进行分支判断，而不是对每个提取结果都一视同仁。

处方代码通常印在背面。

大多数保险公司会在卡的背面打印 BIN、PCN 和团体路由代码。如果这些字段为空，则很可能只提交了正面信息。

该行动被列为 AI-处理健康卡。选择二进制数据后，面板只需要输入二进制字段和健康卡名称，因此文件名中的扩展名决定了格式检测。

健康卡解析与通用解析 OCR

问题	AI-处理健康卡	通用文本提取
What you get back	Named fields such as memberId and rxBin	A block of unstructured text
Layout differences	Handled, fields are identified by meaning	Not handled, position shifts break parsing
Quality signal	Per-field confidence scores	None beyond whether text was found
Work needed after	Map fields and check confidence	Write and maintain regex per insurer
Best for	Cards from many different insurers	One fixed, unchanging document layout

哪些参数 AI- 需要办理健康卡吗？

这三个来源字段互斥。只有与您的匹配项匹配的字段才有效。 输入数据类型 出现在面板中，这就是为什么它们被标记为条件性选项而不是必填选项的原因。

范围	必需的	它的作用	例子
Credential to connect with	Required	The PDF4me account credential the node authenticates with.	PDF4ME account
Action	Required	The action this node performs. Select AI-Process HealthCard.	AI-Process HealthCard
Input Data Type	Required	How the card image reaches the node: binary data from a previous node, a Base64 string, or a URL.	Binary Data
Input Binary Field	Conditional	Name of the binary property holding the card image. Appears when Input Data Type is Binary Data.	data
Base64 Health Card Content	Conditional	The card image encoded as a Base64 string. Appears when Input Data Type is Base64 String.	JVBERi0xLjQKJ...
Health Card URL	Conditional	Publicly reachable URL of the card image. Appears when Input Data Type is URL.	https://example.com/card.jpeg
Health Card Name	Required	Filename of the submitted card including the extension. Drives format detection, so keep it accurate.	health_card.jpeg

输出

该操作返回数据，而非文件。顶级响应字段：

场地	类型	它包含什么
success	Boolean	Whether the extraction completed. Branch on this before reading any field.
message	String	Human-readable status text describing the result.
processedData	Object	Wrapper holding healthData, status, errors, and the job identifiers.
processedData.status	String	Overall processing status for the run.
processedData.errors	Array	Errors encountered during extraction. Empty when the read was clean.
processedData.jobId	String	Unique identifier for the processing job, useful when reporting an issue.
processingTimestamp	String	ISO 8601 timestamp for when processing completed.

里面 processedData.healthData：

场地	类型	它包含什么
memberId	String	The member identification number printed on the card.
memberName	String	Full name of the insured member.
dateOfBirthStr	String	Member date of birth as read from the card.
insuranceProvider	String	Name of the insurer that issued the card.
groupNumber	String	Group or policy number for the plan.
planType	String	Plan category such as PPO, HMO, or EPO.
effectiveDateStr	String	Date the coverage takes effect.
rxBin	String	Prescription BIN routing code, usually printed on the reverse.
rxPcn	String	Prescription processor control number, usually on the reverse.
rxGroup	String	Prescription group identifier, usually on the reverse.
copayInfo	Array	Copay amounts per service type where the card lists them.
confidence	Object	Per-field confidence scores. Use these to gate automatic acceptance.

我该如何设置？ AI-处理健康卡 n8n？

1. 添加 PDF4me 将节点添加到您的工作流程并进行设置 行动 到 AI-处理健康卡。
2. 选择您的 PDF4me 或者创建一个凭证并粘贴您的凭证 API 钥匙。
3. 放 输入数据类型 并将卡片插入匹配字段。
4. 放 健康卡名称 例如，使用正确的扩展名。 health_card.jpeg。
5. 执行节点并检查 processedData.healthData 在输出中。
6. 添加一个 IF 节点来检查 success 以及相关 confidence 得分。
7. 将有把握提取的内容直接发送，其余内容发送到人工审核队列。
8. 仔细决定下游存储哪些数据，只保留流程实际需要的字段。

典型配置

工作流程示例Common n8n workflow patterns using AI-Process HealthCard.

从照片中获取患者信息

1. 新患者通过表格上传病历照片。
2. AI-处理健康卡，读取会员和计划详细信息。
3. 这些字段会预先填写入院记录，供工作人员确认。

正确处理药房索赔

1. 将卡片的背面提交给节点。
2. rxBin、rxPcn 和 rxGroup 作为单独的字段返回。
3. 该索赔申请未经任何人转录代码就直接提交。

置信度控制审查队列

1. 每次提取都会返回每个字段的置信度分数。
2. IF 节点会将它们与您设定的阈值进行比较。
3. 只有不确定的卡片才会送到人手中，因此审核工作量很小。

续保时更新保险范围

1. 续订计划完成后，补发的卡片会陆续寄到。
2. 该节点读取 effectiveDateStr 和 plan 字段。
3. 在旧信息导致申请被拒之前，记录会被更新。

实用技巧

Decide what you store, deliberately

Reading a field does not oblige you to keep it. Persist only what the process needs, and confirm your obligations with whoever owns compliance.

Capture both sides of the card

Member details are on the front and Rx routing codes are usually on the back. Two runs give a complete picture.

Gate on confidence, not just success

A run can succeed while reading a digit wrong. Thresholding the per-field scores is what keeps bad member IDs out of your records.

Test against real insurer variety

Card layouts differ substantially between insurers. Validate on a sample of the cards you genuinely receive, not one example.

Give the model a clean image

Glare, low resolution, and cropped edges cause empty fields far more often than the card layout does.

Keep the filename extension right

Health Card Name drives format detection. A .jpeg image named .pdf will not be handled as you expect.

速查表

场地	价值
Action	AI-Process HealthCard
Input Data Type	Binary Data
Input Binary Field	data
Health Card Name	health_card.jpeg
Credentials	PDF4ME account
Key output path	processedData.healthData
Quality gate	processedData.healthData.confidence

常见问题

What fields does AI-Process HealthCard extract?+

The healthData object carries member identity fields such as memberId, memberName, and dateOfBirthStr, plan fields such as groupNumber, planType, insuranceProvider, and effectiveDateStr, plus the rxBin, rxPcn, and rxGroup prescription routing codes and a copayInfo array. A confidence object gives per-field scores alongside them.

Does using this action make my workflow HIPAA compliant?+

No. Extracting data from a health insurance card does not by itself create or satisfy any compliance position, and PDF4me makes no such claim for this node. The values returned identify a person and their coverage, so they fall under whatever health data obligations already apply to you. In the United States those are set out by the US Department of Health and Human Services; in the EU the GDPR special category rules apply. Confirm your own position with your compliance team before automating this.

What file types can I send to AI-Process HealthCard?+

Card scans and photos in common image formats work, as do PDF captures. Set Health Card Name with the matching extension so the format is detected correctly. Both sides of a card usually need separate runs, since the Rx routing details are typically printed on the back.

How accurate is the extraction and how do I check it?+

The response includes a confidence object with per-field scores, which is the only accuracy signal the action exposes. Use it to route uncertain extractions to human review. Card layouts vary widely between insurers, so validate against a representative sample of the cards you actually receive rather than assuming a fixed accuracy figure.

Why is a field empty in the response?+

A field comes back empty when the value is not present on the side of the card you submitted, or the image is too low in resolution or contrast to read. Rx routing fields in particular are usually on the reverse. Check the errors array and the confidence scores, then re-capture the card at a higher resolution before assuming the card lacks the field.

相关操作

AI处理信用卡

支付卡的读卡方式与支付卡相同，但有其自身的处理义务。

AI 文档解析器

通用解析器，用于解析没有专门操作的文档。

AI-处理银行对账单

另一个敏感文档解析器，可返回结构化字段和置信度。

AI-处理税务文件

对同一接收流程中到达的税务表格进行结构化提取。

在其他平台上执行相同任务

AI-处理健康卡 Make

与……相同的萃取发动机 Make 模块内部 Make 场景。

AI 健康卡解析器 Zapier

从收款表单和数千个 SaaS 应用中触发卡片提取。

AI 健康卡解析器 Power Automate

Microsoft 365 与……流动 SharePoint， Outlook以及 Teams 集成。

获取帮助

n8n Integration Guiden8n SampleGetting Started