Clinical Data Structuring

When structuring clinical data, follow these steps:

1. Extract key information

Identify and extract relevant clinical elements:

• Symptoms and complaints
• Diagnoses or conditions
• Medications (include dosage if available)
• Procedures and treatments
• Laboratory results
• Temporal information (dates, durations, sequence of events)

Only extract explicitly mentioned information.

2. Organize into structured fields

Group extracted information into the following categories:

• symptoms
• diagnoses
• medications
• procedures
• lab_results
• timeline

If a category has no data, return an empty list.

3. Preserve original text

• Keep extracted values as they appear in the input
• Do not normalize or translate
• Do not modify wording

4. Avoid duplication

• Do not repeat the same entity multiple times
• Keep distinct entries separate

5. Handle temporal data

• Represent events with associated time when available
• If time is missing, include "time": "unknown"

6. Output format

Return the result as JSON:

{
  "symptoms": [],
  "diagnoses": [],
  "medications": [],
  "procedures": [],
  "lab_results": [],
  "timeline": [
    {
      "time": "...",
      "event": "..."
    }
  ]
}

7. Constraints

• Do not hallucinate information
• Do not infer missing data
• Only include explicitly stated clinical information
• Keep output strictly in JSON format
• Do not include explanations

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: elisaterumi-ai
• Source: elisaterumi-ai/clinical-agent-skills
• License: Apache-2.0

Install and usage instructions live in the source repository linked above.