Skip to content

niy-ati/AyushDhara

Repository files navigation

AyushDhara AI: Modernizing 5,000 Years of AYUSH Wisdom

AWS Powered Track License: MIT Next.js

Digital Public Infrastructure (DPI) for Inclusive Healthcare. Bridging India's rural health gap through a multilingual, voice-first Digital Vaidya and a real-time Public Health Sentinel dashboard.


The Vision & Problem Statement

AyushDhara AI addresses the Trust Deficit and Accessibility Gap in Indian healthcare. With 70% of India's population in rural areas and practitioners concentrated in urban centers, our solution provides a "Health Guardian" in every pocket, grounded in verified AYUSH scriptures.

📊 Impact Metrics

  • Affordability: Scalable to ₹0.50 - 1.50 per consultation via AWS Serverless.
  • Inclusion: Voice-first interaction in 5+ Indic languages (Hindi, Tamil, Telugu, Bengali, English).
  • Public Impact: Real-time outbreak detection 2 weeks faster than traditional systems.

Technical Architecture

AyushDhara AI utilizes a highly scalable, serverless event-driven architecture designed for the "Next Billion Users."

The Tech Stack

  • AI/LLM: Amazon Bedrock (Claude 3.5 Sonnet) using Retrieval-Augmented Generation (RAG).
  • Voice/NLP: Amazon Transcribe (Indic accents) & Amazon Polly (Natural Indian voices).
  • Data Layer: Amazon DynamoDB (Single-Table Design) & OpenSearch Serverless (Vector Store).
  • Compute: AWS Lambda (Node.js 20.x) & Amazon API Gateway.
  • Security: AWS KMS (PII Encryption) & Bedrock Guardrails.

Core Features

1. Conversational Prakriti Assessment (B2C)

An AI-driven interview to determine a user's Vata/Pitta/Kapha constitution, delivering hyper-personalized preventative care plans based on lifestyle data.

2. Grounded Health Advisory (RAG)

Powered by Amazon Bedrock, ensuring health guidance is strictly retrieved from verified AYUSH datasets to eliminate AI hallucinations.

3. Sentinel Dashboard (B2G)

A public health surveillance engine that anonymizes and aggregates community symptom data to visualize disease hotspots for government officials.


Safety & Compliance (Responsible AI)

To ensure "Direct Selection" in the hackathon, we have prioritized ethics and safety:

  • Emergency Hand-Off: A dedicated lib/safety.ts layer intercepts "Red Flag" symptoms (e.g., chest pain, breathing difficulty) to immediately direct users to 108 Emergency Services.
  • DPDP Act 2023: Implements strict PII anonymization before data enters the Sentinel surveillance loop to protect citizen privacy.
  • ABDM Ready: Data structures utilize FHIR R4 standards and SNOMED CT codes for seamless integration with India's national health mission.

Getting Started

Prerequisites

  • Node.js 20+
  • AWS CLI configured for ap-south-1 (Mumbai)
  • Amazon Bedrock Model Access enabled for Claude 3.5 Sonnet

Setup & Installation

  1. Clone & Install:
    git clone [https://github.com/niyatijain/ayushdhara-ai.git](https://github.com/niyatijain/ayushdhara-ai.git)
    cd ayushdhara-ai
    npm install

About

AyushDhara AI is a voice-first DPI bridging the rural health gap. Using Bedrock RAG, it provides grounded AYUSH guidance and a Sentinel dashboard for proactive, ABDM-ready public health surveillance.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors