Skip to main content

​
Overview

Generate an AI-powered diagnosis prediction for notes associated with a specific appointment in MediSync. This endpoint uses an external AI agent to analyze the clinical notes content and provide diagnostic suggestions with ICD codes and clinical reasoning.
AI-Powered Analysis: This endpoint sends the clinical notes to an external AI diagnosis agent service and processes the response to generate structured diagnostic predictions with ICD codes and clinical reasoning.

​
Authentication

curl -X POST \
  'https://api.medisync.com/notes/diagnosis/apt_abc123?uid=doc_xyz789' \
  -H 'Authorization: Bearer YOUR_JWT_TOKEN'

​
Path Parameters

​
id
string
required
The appointment ID to generate diagnosis prediction for

​
Query Parameters

​
uid
string
required
The user ID of the healthcare provider requesting the diagnosis generation

​
Response

​
success
boolean
Indicates if the operation was successful
​
data
object
Object containing the generated diagnosis prediction
​
error
string
Error message if the operation failed

​
Example Response

{
  "success": true,
  "data": {
    "diagnosis_prediction": {
      "status": "generated",
      "main_diagnosis": "Acute upper respiratory infection",
      "icd_code": "J06.9",
      "reason": "Based on patient symptoms of cough, sore throat, and nasal congestion with no fever, consistent with viral upper respiratory infection"
    }
  }
}

​
Process Flow

  1. Status Update: Initially sets diagnosis prediction status to “pending”
  2. AI Session: Establishes a session with the external AI diagnosis agent
  3. Analysis: Sends clinical notes content to the AI for analysis
  4. Response Processing: Parses and validates the AI response
  5. Database Update: Updates the notes with the generated prediction
  6. Error Handling: Sets status to “failed” with error message if any step fails

​
Behavior Notes

  • Two-Phase AI Interaction: First establishes a session with a “hello” message, then sends the actual clinical content
  • Response Validation: The AI response is parsed and validated for required fields (main_diagnosis, icd_code, reason)
  • Status Tracking: The prediction status is updated throughout the process (pending → generated/failed)
  • Error Recovery: If generation fails, the error is captured and stored in the diagnosis prediction
  • JSON Parsing: The AI response is expected to be in JSON format, with automatic cleanup of markdown code fences

​
Status Codes

  • 200 - Diagnosis prediction generated successfully
  • 400 - Invalid request, parsing error, or AI service failure
  • 404 - Notes not found for the specified appointment