AURA (Intelligent Health Network)

A multimodal AI-driven healthcare platform designed for modern clinical workflows.

The Team

• Nagur Shareef Shaik
• Aasrith Mandava
• Sri Venkata Naga Sai Kakani
• Asish Karthikeya Gogineni

---

🌟 Major Contribution: Smart Oculomics (IRIS)

Our flagship accomplishment is the Smart Oculomics (IRIS) engine. Instead of limiting retinal scans to purely ocular disease classification, we built state-of-the-art predictive models utilizing Vision Transformers (ViT) that empower IRIS to perform profound, systemic health diagnostics.

Key Strength: Diagnosing not just ocular conditions, but predicting Systemic Health and Demographics entirely from a retinal image obtained from a standard mobile device.

By analyzing the micro-vascular structures in the retina, IRIS predicts:

• Demographics: Age & Gender correlation.
• Systemic Health Risks: Diabetes, Hypertension, and overall Cardiovascular Risk.
• Complications: Acute Myocardial Infarction (AMI), Neuropathy, and Nephropathy.

This turns a smartphone into a non-invasive, highly accessible window into a patient’s holistic health profile.

🧭 COMPASS: Agentic RAG System

COMPASS is our dedicated Insurance Guide built as a robust Agentic RAG (Retrieval-Augmented Generation) system.

• It actively retrieves live US health insurance plans, ACA Marketplace data, Medicare/Medicaid eligibility, and geo-localized premiums by proxying our external InsuCompass-API Node.
• It seamlessly assists users who are completely new to health insurance in navigating deductibles, subsidies, and plans, falling back to an on-board Gemini Reasoning engine if the external API is unreachable.

🤖 The AURA Ecosystem of Assistants

Beyond IRIS and COMPASS, AURA features specialized agents acting as healthcare navigators to maintain personal health and encourage preventive care:

• PRISM (Diagnostic): Multimodal diagnostic assistant capable of synthesizing lab reports and clinical documentation.
• SAGE (Wellbeing): A mental health counsellor leveraging Cognitive Behavioural Therapy (CBT) protocols for emotional support.
• APOLLO (Virtual Doctor): Provides accessible symptom consultation, initial triage, and actionable immediate guidance.
• NORA (Dietary Expert): Focuses on preventive care through dynamic, personalized nutritional advice and diet plans.
• VISTA (Visualisation): Generates reports and data visualisations to help users track their systemic health data and biomarkers easily.
• AURA Orchestrator: The platform's master intelligence router. It parses multi-intent user queries and seamlessly delegates them to the correct specialist in the network.

---

📊 Technical Highlights & Benchmarks

To ensure clinical relevance and high accuracy, our predictive Vision Transformer (ViT) models have been assessed and bench-marked against leading healthcare datasets:

• mBRSET: A comprehensive dataset for general retinography and correlating systemic biomarkers.
• MIMIC-CXR: Broad-scale chest radiography reasoning for multimodal alignments.
• DeepEyeNet & ROCO: For extensive ocular abnormalities and general radiology image analysis.

Performance Metrics & Real-World Impact

• Impact: We democratize advanced health screening. By enabling a mobile device to act as a diagnostic capture tool, patients in remote or under-resourced areas can receive systemic health screenings in seconds.
• Outcomes:
  • Substantially reduced diagnosis latency.
  • Significantly lower cost barriers for pre-screening chronic widespread conditions like Diabetes and Hypertension.
  • Empowered decision making: The moment a systemic risk is detected, the AURA ecosystem automatically refers users to preventative care (NORA) or coverage planning (COMPASS), increasing active consumer participation in their own health maintenance.

---

🛠️ Scaffold, Setup & Installation

Tech Stack

• Frontend: React 18, TypeScript, Vite, Custom MNC-grade Vanilla CSS architecture.
• Backend: Python 3.11, FastAPI, Pydantic, Google Gemini 1.5 Pro, ChromaDB (Local persistent Vector Memory per user), SQLite (Relational Data & Threads).

1. Pre-requisites

• Python 3.10+
• Node.js 18+ & npm

2. Environment Configuration

Navigate to the backend directory, duplicate .env.example to .env, and populate it:

# backend/.env
GEMINI_API_KEY=your_gemini_api_key_here

3. Quick Start (Automated Setup)

We provide an interactive start.sh script that handles the entire installation framework for UNIX environments.

From the root directory, grant execution permissions and run:

chmod +x start.sh
./start.sh

What the script does automatically:

1. Backend Validation: Checks for the backend/.venv. If missing, creates the virtual environment and automatically runs pip install -r requirements.txt.
2. Frontend Validation: Checks for frontend-react/node_modules. If missing, runs npm install.
3. Concurrent Launch: Boots the FastAPI backend on http://localhost:8000 and the Vite React UI on http://localhost:5173.

4. Manual Start (Alternative)

Backend:

cd backend
python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
python3 -m uvicorn main:app --reload --port 8000

Frontend:

cd frontend-react
npm install
npm run dev

---

Contact

Email: aura@gmail.com