Healthcare Administration Agents System

An AI-powered multi-agent system for automating medical coding, patient intake triage, and insurance pre-authorization.

Features

• Voice Input Module: Accept audio input and transcribe using Google Speech-to-Text
• Patient Intake Agent: Extract patient details from transcribed text
• Registration Classification: Automatically classify visit types (new/returning/emergency)
• Medical Coding Agent: Auto-assign CPT and ICD-10 codes using RAG
• Insurance Pre-Authorization: Check coverage and authorization requirements
• Real-time Dashboard: Monitor agent processing and view results
• Slack Integration: Automatic notifications for completed intakes

Tech Stack

• Frontend: Next.js 16, React 19, TailwindCSS v4
• Database: PostgreSQL with Prisma ORM
• AI: OpenAI GPT-4, RAG with vector embeddings
• Speech: Google Cloud Speech-to-Text
• External Integration: Slack Web API
• HTTP Client: Axios

Prerequisites

1. Node.js 18+ installed
2. PostgreSQL database running locally
3. OpenAI API key
4. Google Cloud account with Speech-to-Text API enabled
5. Slack workspace with bot token (optional)

Setup Instructions

1. Database Setup

# Install dependencies
npm install

# Set up environment variables
cp .env.example .env
# Edit .env with your actual credentials

# Create database
createdb healthcare_agents

# Run Prisma migrations
npm run prisma:generate
npm run prisma:migrate

# Seed database with sample data
npm run prisma:seed

2. Google Cloud Speech-to-Text Setup

1. Go to Google Cloud Console
2. Create a new project or select existing
3. Enable Speech-to-Text API
4. Create a service account and download JSON credentials
5. Set GOOGLE_APPLICATION_CREDENTIALS in .env to the path of your JSON file

3. OpenAI Setup

1. Get your API key from OpenAI Platform
2. Set OPENAI_API_KEY in .env

4. Slack Setup (Optional)

1. Create a Slack app at api.slack.com/apps
2. Add Bot Token Scopes: chat:write, files:write
3. Install app to workspace
4. Copy Bot User OAuth Token to SLACK_BOT_TOKEN in .env
5. Get channel ID and set SLACK_CHANNEL_ID

5. Run the Application

npm run dev

Visit http://localhost:3000

Usage

1. Upload Audio: Click the microphone icon or drag-drop an audio file
2. Transcription: System automatically transcribes audio using Google Speech-to-Text
3. Agent Processing: Multi-agent workflow processes the intake:
   • Patient information extraction
   • Visit type classification
   • Medical code assignment (CPT/ICD-10)
   • Insurance authorization check
4. View Results: Real-time dashboard shows processing status and results
5. Slack Notification: Automatic notification sent to configured Slack channel

Architecture

Voice Input → Transcription → Multi-Agent Orchestrator
                                      ↓
                    ┌─────────────────┼─────────────────┐
                    ↓                 ↓                 ↓
            Patient Intake    Medical Coding    Insurance Check
                Agent              Agent              Agent
                    ↓                 ↓                 ↓
                    └─────────────────┼─────────────────┘
                                      ↓
                            Dashboard + Slack

Database Schema

• Patient: Patient demographics and insurance info
• IntakeSession: Main workflow tracking table
• MedicalCode: CPT and ICD-10 code assignments
• InsurancePreAuth: Authorization check results
• AgentLog: Detailed execution logs for each agent
• MedicalDocument: RAG knowledge base documents

API Endpoints

• POST /api/intake/transcribe: Upload audio for transcription
• POST /api/intake/process: Process transcribed text through agent workflow
• GET /api/intake/sessions: Get all intake sessions
• GET /api/intake/sessions/:id: Get specific session details
• GET /api/dashboard/stats: Get dashboard statistics

Development

# View database in Prisma Studio
npm run prisma:studio

# Generate Prisma client after schema changes
npm run prisma:generate

# Create new migration
npm run prisma:migrate

License

MIT