🩺 Vitalis AI

Professional Medical Intelligence & Diagnostic Synthesis Platform

Vitalis AI is an advanced, hospital-grade medical intelligence platform designed to transform raw laboratory data and clinical reports into structured, actionable diagnostic insights. Powered by a cascading neural architecture and Retrieval-Augmented Generation (RAG), Vitalis provides healthcare professionals with high-precision medical reasoning and context-aware clinical dialogue.

---

🚀 Key Features

• 🧬 Cascading Neural Architecture: Multi-tier model fallback system (Llama 4 Maverick → Llama 3.3 70B → 8B Instant) ensuring 99.9% diagnostic uptime.
• 📂 Clinical RAG Engine: High-fidelity PDF extraction and semantic search integration for deep-context clinical inquiry.
• 🔐 Identity-Linked Workspaces: Encrypted clinical environments with persistent session telemetry powered by Supabase.
• 📊 Precise Diagnostic Synthesis: Automated analysis of blood reports, pathology, and clinical notes with professional clinical terminology.
• 🩺 Real-time Medical Dialogue: Interactive specialist-grade chat interface for granular exploration of laboratory findings.
• 🎨 Premium Clinical UI/UX: State-of-the-art "Hospital Grade" dark aesthetic with glassmorphic elements and high-density data visualization.

---

🏗️ Architectural Overview

Vitalis AI is structured using a modernized, modular Domain-Driven Design (DDD) to ensure industrial-grade maintainability:

src/
├── core/                # Clinical Intelligence Logic
│   ├── agents/          # Neural Agents (Analysis, RAG, Model Managers)
│   ├── services/        # Orchestration layers (AI, Session)
│   └── workspace_manager.py
├── ui/                  # Frontend Interface
│   ├── components/      # Branded Atomic Components (Auth, Forms, Sidebar)
│   ├── pages/           # High-level Router and Interface Logic
│   └── widgets/         # Composite Interaction Blocks
├── infra/               # System Infrastructure
│   ├── identity_service.py  # Supabase Auth Integration
│   └── document_service.py  # Clinical OCR/Extraction logic
├── config/              # Centralized System Settings & Prompts
└── utils/               # Validations, PDF Extractions & Helpers

---

🛠️ Technology Stack

• Logic Core: Python 3.10+
• UI Framework: Streamlit (Modernized with Custom CSS Overrides)
• Neural Backend: Groq Inference Engine (Llama 3.1/3.3/4)
• Vector Space: LangChain + FAISS + HuggingFace Embeddings
• Identity & Persistence: Supabase (PostgreSQL + Auth)
• Clinical OCR: PDFPlumber + FileType

---

🚥 Quick Start

1. Prerequisite Configuration

Create a .streamlit/secrets.toml file in the root directory:

GROQ_API_KEY = "your_groq_api_key_here"
SUPABASE_URL = "your_supabase_url_here"
SUPABASE_KEY = "your_supabase_anon_key_here"

2. Dependency Installation

pip install -r requirements.txt

3. Initialize Neural Core

streamlit run src/main.py

---

🛡️ Security & Privacy

Vitalis AI is built with clinical-grade safety in mind:

• Ephemeral Memory: Clinical documents are processed in-memory and NOT used for model training.
• Encrypted Transport: All telemetry is communicated over TLS-encrypted channels to Supabase.
• Role-Based Access: Multi-tenant workspace isolation ensures data integrity.

---

📄 License & Attribution

This project is created and maintained by Rajkaran. Licensed under the MIT License. See LICENSE for details.

---

Bridging the Gap Between Clinical Data and Actionable Intelligence.
Built with ❤️ for the future of Healthcare.