Arogya AI

Arogya AI is an AI-powered Radiology Triage & Diagnostic Intelligence Platform. It streamlines hospital workflows by automatically analyzing patient scans (X-rays, CT Scans, Brain MRIs), combining image anomalies with clinical lab notes, and presenting a prioritized worklist to radiologists.

🚀 Basic Workflow

1. Patient Intake: Doctors register patients and attach clinical notes and lab records via the Hospital Staff Portal.
2. Scan Upload: Technologists select pending patients and upload their diagnostic scans (X-ray, CT, MRI).
3. AI Analysis (FastAPI Backend):
   • Scans are processed by specialized ML models (e.g., Pneumonia X-ray, CT Scan, Brain MRI endpoints).
   • The AI correlates visual image anomalies with clinical lab severity to generate critical_flags, an image_score, and a final_priority_score.
4. Radiologist Triage: Radiologists receive an automatically sorted smart worklist. They review the AI's explanation (heatmap, image findings, lab correlation).
5. Approval & Feedback: Radiologists approve the AI prediction or reject it with corrected priorities (feeding back into the model).
6. Doctor Results: Doctors review the finalized reports, sorted by urgency.
7. Patient Portal: Patients can log in using their ID to track their scan status, view AI predictions, and access final reports.

💻 Tech Stack

Frontend (/frontend)

• Framework: React 18 with Vite
• Language: TypeScript
• Styling: Tailwind CSS
• UI Components: Shadcn UI, Radix UI
• Animations: Framer Motion
• Icons: Lucide React
• Routing: React Router

Database & Node Backend (/server)

• Runtime: Node.js
• Framework: Express.js
• Database: MongoDB (using Mongoose for schemas: Patient, DoneReport)
• Features: Patient state tracking, archived report storage.

AI Engine (/backend)

• Framework: FastAPI (Python)
• Computer Vision & NLP: PyTorch / OpenCV (MedSigLIP and classification models)
• Endpoints:
  • POST /api/pneumonia-xray
  • POST /api/ct-scan
  • POST /api/brain-mri
• Features: Multi-modal analysis parsing lab notes and generating combined severity scores.

🛠️ Getting Started

Prerequisites

• Node.js (v18+)
• Python 3.9+
• MongoDB instance running locally (port 27017)

Running the Full Stack

This project is configured with a root package.json to concurrently start the Node/React components.

1. Install dependencies for both frontend and node server:

   npm run install:all

2. Start the Frontend and Node API:

   npm run dev

   Frontend gracefully runs on localhost:8080

3. Start the Python AI Backend: Open a new terminal, navigate to the backend folder, and run:

   pip install -r requirements.txt
   uvicorn main:app --host 0.0.0.0 --port 8000 --reload