Team 404 Not found 1.Srikanth 2.Neeraj 3.Pavan 4.Abilash
The "VitalGrid" Master Code Generation Prompt Role: You are a Senior Full-Stack Engineer and AI Architect specializing in Health-Tech and Privacy-Preserving AI.
Project Goal: Build a functional PoC for VitalGrid, a decentralized health network that uses Edge AI and Gemini to analyze medical data without compromising privacy.
Tech Stack: - Frontend: React.js with Tailwind CSS (Modern, clean, "medical" UI).
Backend: Python (FastAPI) to handle Gemini API orchestration.
AI Integration: Google Gemini API (for PII stripping and Clinical Structuring).
Features to Simulate: Multimodal inputs (Acoustic/Vision), Digital Twin baseline, and Federated Learning updates.
Please generate the code for the following components:
- The "Privacy Shield" Orchestrator (Python/FastAPI) Create a backend script that receives raw, sensitive patient data. It must use the Gemini API with a specific System Instruction to:
Identify and redact all PII (Name, DOB, Location).
Perform "Clinical Structuring": Convert messy symptoms into a standardized JSON object (ICD-10 codes, severity score).
Generate "Synthetic Clinical Data": Create a plausible, non-identifiable version of the case for the Research Cloud.
Generate a "Zero-Knowledge Proof" (mocked as a secure hash/token) to verify the data's integrity without revealing identity.
- The Patient Edge App (React/Tailwind) Create a high-fidelity dashboard with three main views:
Triage Center: Includes a text input for symptoms and "Mock" buttons for Multimodal Diagnostics (e.g., "Analyze Cough Sound" using Web Audio API and "Scan Skin Lesion" using Camera).
Digital Twin Dashboard: A visual representation of the user’s health baseline. Show a "Normal Baseline" graph vs. "Current Deviation" based on the input symptoms.
The "Privacy Wall" Animation: A visual transition showing the raw data hitting a "Gemini Shield" and transforming into anonymized, glowing green JSON packets before being sent to the cloud.
- The Decentralized Research Feed (React) A separate view (representing the "Global Immune System") that displays a live feed of incoming Synthetic Data from various nodes.
Show a map marking "Anonymized Signal Clusters" (e.g., "5 cases of respiratory distress detected in North Sector").
Ensure no personal names are visible, only clinical IDs and severity levels.
- The Federated Update Logic (Javascript/Python) Include a function that simulates Federated Learning. Instead of sending the patient's data, it should calculate a "Model Weight Update" (a JSON of learned patterns) and send that to the central server.
Specific Implementation Requirements:
System Prompt for Gemini: "You are a Medical Privacy Guard. Your task is to receive raw patient input, strip all PII, and output ONLY a structured JSON containing: 1. Standardized Medical Condition, 2. Severity (1-10), 3. Required Specialist, and 4. A Synthetic Summary for research. DO NOT return names or dates."
Visual Style: Use a "Dark Mode" medical aesthetic with teal and electric blue accents (representing the 'Grid').
Mock Data: Include a mock_database.js with verified doctors and their blockchain-stored credentials.
Deliverable: Provide the complete, modular code for main.py (Backend), App.js (Frontend), and GeminiOrchestrator.js (AI Logic).
How to use this prompt: Get your Gemini API Key from the Google AI Studio.
Paste the prompt above into your preferred AI coding assistant.
Run the Backend: Once the code is generated, run the FastAPI server.
Run the Frontend: Launch the React app to see the "Privacy Wall" in action.
Why this wins the Hackathon: Technical Sophistication: You aren't just calling an API; you are demonstrating Edge AI and Data Synthesis.
Visual Storytelling: The "Privacy Wall" animation makes the abstract concept of "PII stripping" tangible for the judges.
Addressing Feedback: This code explicitly includes the Multimodal and Digital Twin features that the judges felt were missing from standard health apps.