The Healthcare AI Ecosystem

Two Integrated Solutions for Complete Health Protection

Integrating individual health intelligence with population-level disease surveillance — the only platform of its kind.

Overview

This ecosystem addresses two parallel crises in global healthcare:

Problem	Scale
Underserved populations without healthcare access	80% of rural populations
Disease outbreaks reported manually with delays	Hours to days lag time
Language barriers blocking effective diagnosis	Billions affected globally
No centralized real-time disease surveillance	Every country lacks unified data

Our solution: Two purpose-built platforms — SYMPTOMAP for population-level disease surveillance, and the Health Agent for individual AI-powered health consultation — sharing a real-time data bridge so that when a doctor reports an outbreak, affected individuals are warned instantly.

The Two Platforms

SYMPTOMAP — Real-Time Disease Surveillance

Live App: symtomap-main.vercel.app | Backend: symptomap-2-python.vercel.app

SYMPTOMAP is an interactive, map-based disease surveillance platform for healthcare professionals. Doctors can report outbreaks in under 30 seconds; administrators get a live dashboard of disease spread across India.

Core Capabilities:

Interactive MapLibre Map — visualize outbreaks geographically with severity heatmaps
30-Second Submission — doctor portal for rapid outbreak reporting
Live Admin Dashboard — real-time stats: 200,000+ seeded records across India
WebSocket Alerts — instant broadcast to all connected users on new reports
Doctor Station — secure authenticated portal for submissions and alert creation
Admin Console — approval workflows, broadcast system, verification pipeline
Analytics Engine — trend analysis, week-over-week comparisons, activity feeds
Role-Based Access — Admin, Doctor, and Public tiers with JWT authentication
Air Quality Intelligence — Urban AQI monitoring, epidemiological zoning, and viral fever risk prediction

Health Agent — AI-Powered Health Consultation

App: healthzy.app | Source: github.com/Rajkaran-122/Health_agent

The Health Agent is a 24/7 AI doctor accessible from any device, in any language. It accepts images, voice, and text to conduct an intelligent clinical interview and generate a diagnostic report with confidence scores.

Core Capabilities:

Image-Based Disease Detection — multi-model ensemble with 85%+ accuracy
Voice-Activated Consultation — real-time voice via LiveKit WebRTC (<200ms latency)
Multilingual First — designed for global accessibility, not English-centric
Multi-LLM Failover — OpenAI + Claude + Grok with 99.9% uptime
SOAP Note Generation — clinical-format diagnostic reports with confidence scores
SYMPTOMAP Integration — receives real-time outbreak alerts from SYMPTOMAP
2G Compatible — works on low-end devices in underserved areas
Personal Health Records — encrypted, user-controlled health history

How the Ecosystem Works

%%{init: {'theme': 'dark', 'themeVariables': { 'primaryColor': '#1e40af', 'primaryTextColor': '#ffffff', 'lineColor': '#60a5fa', 'mainBkg': '#0f172a', 'textColor': '#e2e8f0'}}}%%
flowchart LR
    subgraph Individual["Individual Layer"]
        User["Patient / Public"]
        HA["Health Agent\nhealthzy.app"]
        User -->|"Voice / Image / Text"| HA
    end

    subgraph Professional["Professional Layer"]
        Doctor["Doctor / PHO"]
        SM["SYMPTOMAP\nsymtomap-main.vercel.app"]
        Doctor -->|"30-sec Submission"| SM
    end

    subgraph Intelligence["Intelligence Layer"]
        DB[("200K+ Records\nSQLite + PostGIS")]
        WS["WebSocket\nReal-Time Bridge"]
        AI["Multi-LLM\nOrchestration"]
        SM -->|"Writes outbreak data"| DB
        DB -->|"Feeds analytics"| WS
        HA -->|"Requests diagnosis"| AI
        AI -->|"SOAP Report"| HA
    end

    subgraph AlertLayer["Alert Layer"]
        Broadcast["Broadcast System\nSMS / Push / WebSocket"]
        WS -->|"Outbreak in area"| Broadcast
        Broadcast -->|"Outbreak warning"| User
    end

    style Individual fill:#1e3a5f,stroke:#3b82f6
    style Professional fill:#14532d,stroke:#22c55e
    style Intelligence fill:#3b1f5e,stroke:#a855f7
    style AlertLayer fill:#7c2020,stroke:#ef4444

The Integration Flow:

A doctor spots an outbreak and submits via SYMPTOMAP in 30 seconds
The submission hits the FastAPI backend, is stored, and a WebSocket event fires
The AI analytics engine zones the affected area and calculates risk levels
An instant alert is dispatched to Health Agent users in that geographic area
Individuals get early warnings and personalized prevention guidance

Architecture

SYMPTOMAP Architecture

%%{init: {'theme': 'dark', 'themeVariables': {'primaryColor': '#1e40af', 'lineColor': '#60a5fa', 'mainBkg': '#0f172a', 'textColor': '#e2e8f0'}}}%%
graph TD
    subgraph Frontend["Frontend — React 18 + TypeScript"]
        MapUI["MapLibre GL Map\nOutbreak Heatmap"]
        AdminUI["Admin Console\nApproval + Broadcasts"]
        DoctorUI["Doctor Station\nReport Submission"]
        PublicUI["Public Map\nDisease Tracker"]
    end

    subgraph Backend["Backend — FastAPI Python Async"]
        Auth["JWT Auth\n24hr Sessions"]
        OutbreakAPI["Outbreak API\nCRUD + Filtering"]
        StatsAPI["Stats API\nAggregated Metrics"]
        AdminAPI["Admin API\nApproval Workflow"]
        WSAPI["WebSocket\nReal-Time Events"]
        BroadcastAPI["Broadcast API\nHealth Advisories"]
        AirQualityAPI["AQI Intelligence\nUrban Air Quality"]
    end

    subgraph DataLayer["Data Layer"]
        SQLite[("SQLite\n200,002 Records")]
        MockRedis["Redis Pub-Sub + Cache"]
        AuditLog["Audit Log\nAll Actions Tracked"]
    end

    subgraph AILayer["AI Layer — Celery Workers"]
        Summarizer["Outbreak Summarizer\nAI Analysis"]
        Triage["Triage Agent\nSeverity Scoring"]
        ZoneAgent["Zoning Agent\nEpidemiological Zones"]
        AQIAgent["AQI Agent\nViral Fever Prediction"]
    end

    MapUI & AdminUI & DoctorUI & PublicUI --> Auth
    Auth --> OutbreakAPI & StatsAPI & AdminAPI & BroadcastAPI & AirQualityAPI
    OutbreakAPI & StatsAPI & AdminAPI --> SQLite
    WSAPI --> MockRedis
    OutbreakAPI --> Summarizer & Triage & ZoneAgent & AQIAgent
    Summarizer & Triage & ZoneAgent & AQIAgent --> SQLite
    MockRedis --> WSAPI
    AuditLog --> SQLite

Health Agent Architecture

%%{init: {'theme': 'dark', 'themeVariables': {'primaryColor': '#7c3aed', 'lineColor': '#a78bfa', 'mainBkg': '#0f172a', 'textColor': '#e2e8f0'}}}%%
graph TD
    subgraph Input["Input Layer"]
        Voice["Voice Input\nLiveKit WebRTC"]
        Image["Image Upload\nDisease Photos"]
        Text["Text / Chat\nSymptom Description"]
    end

    subgraph Orchestration["LangGraph Orchestration"]
        Intake["Patient Intake Agent\nAge, Gender, Timeline"]
        Interview["Clinical Interview Agent\nAdaptive Questioning"]
        Scoring["Risk Scoring Agent\nSymptom Weighting"]
        Differential["Differential Diagnosis Agent\nTop-3 Possibilities"]
        Report["SOAP Note Generator\nStructured Output"]
    end

    subgraph LLMs["Multi-LLM Layer — Failover"]
        GPT["OpenAI GPT-4o\nPrimary"]
        Claude["Anthropic Claude\nFallback 1"]
        Grok["xAI Grok\nFallback 2"]
    end

    subgraph Output["Output Layer"]
        DiagReport["Diagnostic Report\nConfidence Scores"]
        Recs["Recommendations\nActionable Advice"]
        Translated["Multilingual Alert\nHindi, Tamil, etc."]
        OutbreakWarn["Outbreak Warning\nFrom SYMPTOMAP"]
    end

    Voice & Image & Text --> Intake
    Intake --> Interview --> Scoring --> Differential --> Report
    Interview & Scoring & Differential <-->|"LLM Calls with Failover"| GPT
    GPT -->|"Fails"| Claude
    Claude -->|"Fails"| Grok
    Report --> DiagReport & Recs & Translated
    OutbreakWarn -.->|"SYMPTOMAP Bridge"| Recs

AI Doctor Clinical Flow

%%{init: {'theme': 'dark', 'themeVariables': {'primaryColor': '#065f46', 'lineColor': '#34d399', 'mainBkg': '#0f172a', 'textColor': '#e2e8f0'}}}%%
flowchart TD
    Start(["First Step"]) --> Symptoms["Enter Symptoms"]
    Symptoms --> Demographics["Enter Age and Gender"]
    Demographics --> Timeline["When symptoms started"]
    Timeline --> RiskScore["AI Score generated\nrelevant questions with variables"]
    RiskScore --> Findings["Store Findings Till Now\nbased on User Answers"]
    Findings --> Discussion{"Discussion Point"}

    Discussion -->|"Yes/No — No Clear Type of Questions"| OpenQuestions["Open Ended Questions"]
    Discussion -->|"More Questions Needed"| MoreQ["Generate More Questions"]

    RiskScore --> AINarrow["AI narrows and notes\nanalysis at step 5"]
    AINarrow --> StartTaking["Start Taking down history\nwith focused questions"]

    StartTaking --> MoreDetails{"More Details about\ntest or analysis needed"}
    MoreDetails -->|"Yes"| HistoryCollection{"History\nCollection Completed"}
    MoreDetails -->|"No"| HistoryCollection

    HistoryCollection -->|"Yes"| AskSymptom["Ask what makes symptoms\nbetter or worse + follow-up"]
    AskSymptom -->|"Yes"| AskFollowUp["Ask additional questions\nfrom previous step"]
    AskFollowUp -->|"No"| AskOther["Ask for any other symptoms"]

    AskOther -->|"Yes"| AnyAdditional{"Any Additional Symptoms"}
    AnyAdditional -->|"Yes"| AskMoreSimilar["Ask if symptom was\nreported in previous step"]
    AnyAdditional -->|"No"| FullDetails["Full Details about\nadditional symptoms"]

    AskMoreSimilar -->|"Yes"| FullHistory["Full Details about\nhistory of reported symptom"]

    StartTaking --> AskMeds["Ask about Medications,\nSupplements, etc."]
    AskMeds --> MedDetails{"Full medication\ndetails available"}
    MedDetails -->|"No"| Allergies{"Any Allergies or\nother relevant info"}
    MedDetails -->|"Yes"| Allergies

    Allergies -->|"Yes"| AllergyDetails["Full Details allergies\nor additional info"]
    Allergies -->|"No"| AskMoreNeeded["Ask More Questions\nif needed"]

    AllergyDetails -->|"Yes"| Consult["Generate Consultation Summary\nDiagnostic Report + Next Steps"]
    AskMoreNeeded -->|"Yes"| Consult
    FullHistory --> Consult
    FullDetails --> Consult

    Consult --> End(["Complete Report + SOAP Note"])

    style Start fill:#065f46,color:#fff
    style End fill:#065f46,color:#fff
    style Consult fill:#1e40af,color:#fff
    style Discussion fill:#7c3aed,color:#fff
    style AnyAdditional fill:#7c3aed,color:#fff
    style MedDetails fill:#7c3aed,color:#fff
    style Allergies fill:#7c3aed,color:#fff
    style MoreDetails fill:#7c3aed,color:#fff
    style HistoryCollection fill:#7c3aed,color:#fff

Notes: SOAP findings are updated in real-time after each step. AI also sends reasoning behind every response.

Key Features

SYMPTOMAP Features

Feature	Description
Interactive Map	MapLibre GL with outbreak markers, severity heatmaps, and zone boundaries
30-Second Submission	Doctor submits disease, severity, location — complete in seconds
Live Dashboard	Total reports, pending review, high priority, active cases — all real-time
Admin Console	Approve/reject outbreak reports, manage broadcasts, verify submissions
Broadcast System	Create and send public health advisories with urgency levels
WebSocket Real-Time	All clients receive instant updates without page refresh
Analytics	Activity feed, week-over-week trends, disease distribution charts
Air Quality Module	AQI monitoring, viral fever risk prediction, epidemiological zoning
Security	JWT auth, role-based access, audit logs, input sanitization, rate limiting
Approval Workflow	Pending to Under Review to Approved/Rejected pipeline

Health Agent Features

Feature	Description
Voice Consultation	Real-time voice via LiveKit WebRTC with sub-200ms latency
Image Diagnosis	Upload photos for AI-powered disease detection — 85%+ accuracy
Multilingual	Supports Hindi, Tamil, Bengali, and more regional languages
Multi-LLM	OpenAI + Claude + Grok with automatic failover for 99.9% uptime
SOAP Reports	Clinical-format reports with confidence scores, not just chatbot answers
Outbreak Alerts	Receives real-time warnings from SYMPTOMAP for affected areas
Low-End Compatible	Works on basic 2G phones — designed for underserved populations
Health Records	Personal health history management with encryption

Technology Stack

SYMPTOMAP

Layer	Technologies
Frontend	React 18+, TypeScript, Vite, MapLibre GL JS, Leaflet
Backend	FastAPI (Python async), SQLAlchemy, Pydantic v2
Database	SQLite (200K+ records), Redis Pub/Sub, PostgreSQL-ready
Auth	JWT tokens (24-hour sessions), bcrypt password hashing
Realtime	WebSockets, Mock Redis (prod: Redis Pub/Sub)
Security	Input sanitization, CORS enforcement, rate limiting, audit logging

Health Agent

Layer	Technologies
Frontend	Next.js (React), TypeScript
Backend	Flask microservices, Python
AI/ML	OpenAI GPT-4o, Anthropic Claude, xAI Grok (multi-LLM failover)
Voice	LiveKit WebRTC (<200ms latency)
Orchestration	LangGraph (stateful conversation flows)
Database	MongoDB, Redis (60-70% reduced load)

Live Demos and Links

Platform	Link	Description
SYMPTOMAP Main	symtomap-main.vercel.app	Primary frontend — full ecosystem entry point
SYMPTOMAP Python API	symptomap-2-python.vercel.app	FastAPI backend with live Swagger UI
Healthzy Health App	healthzy.app	AI health assistant — live app
Health Agent Source	github.com/Rajkaran-122/Health_agent	AI agent source code and architecture

Pilot Results

SYMPTOMAP Pilot — 50 Doctors, 2 Months

Metric	Result
First-Week Adoption	94%
Avg. Submission Time	30 seconds
Data Accuracy	98%
Faster than Phone/Fax	89% faster

Zero data loss. 3 outbreak clusters detected and contained early.

Health Agent Pilot — 100 Users, 3 Months

Metric	Result
Daily Active Users	87%
Satisfaction Rating	92%
Early Issue Detection	73%
Serious Conditions Caught Early	12+

Combined Impact: Early detection and containment of 3 outbreak clusters, preventing over 50 cases through early intervention and warnings to Health Agent users.

Roadmap

Phase 1 — Months 1 to 6

Deploy in 3 to 5 pilot regions across India
Onboard 1,000 users and 500 doctors
Add 7 more languages (total 10 supported)
Email/SMS notification system
Mobile PWA launch

Phase 2 — Months 7 to 12

Scale to 50,000 users
ML-based outbreak prediction engine
Hospital system API integration
Native mobile apps (iOS/Android)
Advanced analytics dashboard

Phase 3 — Year 2 and beyond

Documentation Directory

The docs/ folder contains the full technical and product documentation for this ecosystem. Documents are organized by layer — architecture, product, implementation, and legal — and together form a complete picture of how the system is designed, built, and operated.

Architecture Documents

High-Level Design — HLD

The authoritative system architecture document. Contains the full Mermaid architecture diagram covering all data sources (doctor submissions, IoT AQI sensors, satellite/meteorological APIs, public health records), the FastAPI ingestion gateway, the SQLite/PostgreSQL persistence layer, the Redis Pub/Sub messaging backbone, and the Celery multi-agent AI worker cluster — comprising the Outbreak Summarizer, Triage Agent, Epidemiological Zoning Agent, and AQI Intelligence Agent. Documents all client-facing surfaces: Admin Command Center, Doctor Station, Public Map, and the Health Agent integration bridge. Covers the extended Air Quality Intelligence module and key architectural decisions including geography column deferral for SQLite/PostGIS dual compatibility.

Low-Level Design — LLD

The micro-level specification. Contains three complete sequence diagrams: (1) Doctor Outbreak Submission and Admin Approval pipeline — the full lifecycle from authenticated POST through Celery AI processing to WebSocket broadcast and map update; (2) Health Agent Consultation Flow — showing LangGraph orchestrating multi-LLM failover and injecting real-time SymptoMap outbreak context into a clinical interview session; (3) Entity-Relationship Diagram covering all eight production database tables (users, hospitals, outbreaks, doctor_outbreaks, predictions, zones, alerts, aqi_signals) with column-level detail. Includes full API contract specifications for all major endpoints with request/response schemas.

System Architecture — V1 Analysis and V2 Target Design

A rigorous technical analysis comparing the current V1 implementation against the target V2 enterprise architecture. Identifies the four V1 bottlenecks: the 30-second polling anti-pattern generating 20,000 wasted requests per minute under load; the SQLite write-lock problem causing data loss under concurrent doctor submissions; monolithic read/write coupling causing dashboard reads to starve outbreak write paths; and absence of predictive intelligence reducing the system to a CRUD map. Documents the V2 architectural responses: CQRS with Kafka/Redis Streams event queues for zero-loss ingestion, WebSocket push replacing polling, PostGIS spatial indexing for millisecond geographic queries, ML microservice decoupling via Ray, and multi-tier Redis caching for geospatial tile re-use. Includes tech stack upgrade matrix and failure mode analysis with Kubernetes HPA scaling.

Product Documents

Business Requirements Document — BRD

The complete product specification. Covers executive summary, problem statement, full user persona analysis (public health officers, doctors, administrators, individual patients), functional and non-functional requirements, user story specifications with acceptance criteria, competitive positioning matrix against Teladoc, ProMED, HealthMap, and CDC systems, business model, revenue projections (Year 1: $50K, Year 2: $500K, Year 3: $2M), partnership strategy, and the three-phase strategic roadmap.

Doctor Station — BRD

The product specification for the Doctor Station portal specifically. Covers the secure authentication flow (shared station password plus JWT), the outbreak submission form design rationale for the 30-second submission target, alert creation workflow, submission history display, and the approval state machine. Includes UI/UX specifications and error handling requirements for all submission edge cases.

API Specification

All public and authenticated API endpoint contracts. Covers authentication requirements, request/response schemas, HTTP status codes, and rate limiting behavior for every route across the outbreak, doctor, admin, stats, analytics, broadcast, prediction, and WebSocket namespaces. Includes the Swagger UI available at the live backend endpoint.

Database Schema

Complete database schema documentation with entity definitions, column types, constraints, indexes, and foreign key relationships. Documents the dual-table design rationale — the ORM outbreaks table for structured incoming data and the doctor_outbreaks table holding the 200,000+ seeded bulk records — and explains the SQLAlchemy deferred() wrapper on Geography columns to prevent SQLite AsEWKB/GeomFromEWKT crashes while preserving PostGIS compatibility for production deployment.

Operations Documents

Doctor User Guide

Step-by-step operational guide for healthcare providers using the Doctor Station portal. Covers login, outbreak submission workflow (disease type selection, severity classification, geographic location marking, patient count entry), alert creation, and submission history review.

Deployment Guide

Full production deployment instructions. Covers environment variable configuration, database migration steps, Redis setup, Celery worker launch, frontend build and Vercel deployment, ASGI server setup with Gunicorn plus Uvicorn workers, CORS configuration for production domains, and HTTPS termination. Covers both the Vercel-hosted frontend path and the self-hosted backend path.

Legal

Proprietary Notice

Intellectual property terms, restrictions on reproduction and commercial use, and contact information for licensing and partnership inquiries.

Setup and Running

Prerequisites

Node.js 18+, Python 3.10+, npm / pip

SYMPTOMAP — Local Development

# Clone the repository
git clone https://github.com/Rajkaran-122/sympto-pulse-map-main
cd sympto-pulse-map-main

# Backend (FastAPI)
cd backend-python
python -m venv venv
venv\Scripts\activate          # Windows
source venv/bin/activate       # Mac/Linux
pip install -r requirements.txt
python -m uvicorn app.main:app --host 0.0.0.0 --port 8000

# Frontend (React + Vite) — in a new terminal
cd frontend
npm install
npm run dev

Default Credentials:

Role	Email	Password
Admin	admin@symptomap.com	Admin@123
Doctor	doctor@symptomap.com	Doctor@123

Frontend: http://localhost:5173 · Backend API: http://localhost:8000 · Swagger UI: http://localhost:8000/docs

Security and Ethics

Security Architecture

HIPAA-Compliant architecture with end-to-end encryption
JWT authentication with 24-hour auto-expiry
Protection against SQL injection, XSS, CORS enforcement
Rate limiting (100 req/min), full audit logging, automated backups
Role-based access control — Admin, Doctor, Public tiers

Ethical Principles

Privacy First — User data controlled by the user, never sold
AI Assists, Does Not Replace — Transparent confidence scores on every diagnosis
Equity and Inclusion — Free tier, multilingual, 2G-compatible for underserved populations
Transparency — Open-source core, explainable AI outputs, regular audits
Social Impact — Free deployment for government/NGOs; anonymized data for medical research only with consent

Our Goal: 30% reduction in preventable deaths by 2030.

License

This repository and all its contents are Proprietary and Confidential.

Unauthorized reproduction, distribution, or commercial use is strictly prohibited. See docs/PROPRIETARY_NOTICE.md for full terms. For commercial licensing, government deployment, or partnership opportunities — contact directly.

Contact

Rajkaran Yadav — Team Lead, Full-Stack Developer, Healthcare AI Systems Specialist

Phone: +91-9975889977
GitHub: github.com/Rajkaran-122
Live App: symtomap-main.vercel.app

Key Skills: Frontend (React, Next.js) · Backend (Python, Node.js) · AI/ML · Databases · System Architecture

"Every person has instant access to quality healthcare. Every outbreak is detected before becoming an epidemic."

Name		Name	Last commit message	Last commit date
Latest commit History 170 Commits
.github/workflows		.github/workflows
backend-python		backend-python
data		data
docs		docs
flask-otp-auth		flask-otp-auth
frontend		frontend
legacy_frontend_config		legacy_frontend_config
public		public
scripts		scripts
src		src
.env		.env
.env.example		.env.example
.gitignore		.gitignore
LICENSE		LICENSE
README.md		README.md
SECURITY.md		SECURITY.md
docker-compose.yml		docker-compose.yml
pyrightconfig.json		pyrightconfig.json
render.yaml		render.yaml
start.bat		start.bat
start.sh		start.sh
version.json		version.json

Folders and files

Latest commit

History

Repository files navigation