[FEATURE]: Structured AI Medical Insights for Transcription Pipeline

[Shweta-281]

Feature and its Use Cases

Overview

Currently, the transcription pipeline generates unstructured text outputs for summary and prescription, which limits scalability and real-world usability.

Problem

• AI responses are plain text
• No structured medical data extraction
• Difficult to extend for future features (EHR, analytics, diagnosis)

Proposed Solution

Introduce a structured AI response system using a new MedicalInsights model.

Features

• JSON-based AI output (summary, symptoms, medicines)
• New domain model: MedicalInsights
• Improved Gemini service for structured parsing
• Controller updated to handle structured data
• UI ready for rendering structured insights

Architecture

Audio → Deepgram → Transcript
→ Gemini AI (JSON output)
→ MedicalInsights Model
→ Controller → UI

Impact

• Makes the app closer to real-world healthcare use
• Enables future extensions (diagnosis, analytics)
• Improves maintainability and scalability

Tasks

• Create MedicalInsights model
• Update GeminiService
• Update TranscriptionController
• Update UI screens

Additional Context

No response

Code of Conduct

• I have joined the Discord server and will post updates there
• I have searched existing issues to avoid duplicates