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.
Log in or sign up for Devpost to join the conversation.