HealthHub AI 🩺

⚠️ Note: This application is currently under active development. Features and endpoints may change frequently as we continue building out the core platform.

Welcome to the HealthHub AI project! This repository contains the foundation for an intelligent Triage and Doctor Marketplace platform, designed to accurately evaluate patient symptoms using AI and connect them with the right specialists.

Tech Stack

Frontend

Backend & Database

AI & Infrastructure

Platform Overview

HealthHub AI is an AI-powered medical triage and telehealth platform designed to accurately evaluate patient symptoms and connect them with verified specialists.

🚀 Key Capabilities & Achievements

• Architecture & Full-Stack: Engineered a scalable platform using React, FastAPI, and PostgreSQL, featuring 30+ RESTful API endpoints for robust patient-doctor workflows.
• AI & Vector Search: Architected a Retrieval-Augmented Generation (RAG) system with pgvector, embedding 50,000+ medical knowledge chunks to achieve <200ms semantic search latency for real-time symptom triage.
• Intelligent Triage Engine: Developed an advanced NLP pipeline utilizing LLMs (Groq/Gemini/OpenAI) to extract symptoms, detect emergency red flags, and calculate automated medical risk assessments.
• Real-Time Telehealth: Integrated end-to-end encrypted WebRTC video consultations with PeerJS, ensuring secure, HIPAA-compliant telehealth appointments.
• Infrastructure & Deployment: Streamlined application delivery by fully containerizing the architecture with Docker and Docker Compose, reducing local setup time and ensuring environment consistency.
• Enterprise Security: Implemented robust security using JWT authentication and Role-Based Access Control (RBAC) to securely manage Patient, Doctor, and Admin interfaces.

System Architecture

graph TD
    %% Frontend Layer
    subgraph Frontend ["React / Vite Frontend"]
        UI["User Interface"]
        PatientDash["Patient Dashboard"]
        DoctorDash["Doctor Dashboard"]
        AdminDash["Admin Panel"]
    end

    %% Backend Layer
    subgraph Backend ["FastAPI Backend"]
        API["API Router"]
        AuthSvc["Auth & RBAC Service"]
        DocSvc["Doctor Onboarding & PDF Service"]
        
        %% Triage Pipeline
        subgraph TriagePipeline ["Intelligent Triage Pipeline"]
            SympExt["Symptom Extraction (LLM)"]
            RedFlag["Red-Flag Emergency Detection"]
            TriageScore["Triage Scoring Engine"]
        end
        
        %% RAG System
        subgraph KnowledgeSystem ["RAG Knowledge System"]
            CorpusPipe["Corpus Ingestion Pipeline"]
            RAG["RAG Retrieval Service"]
        end
    end

    %% Database Layer
    subgraph Database ["PostgreSQL + pgvector"]
        UsersDB[("Users & Roles")]
        ProfilesDB[("Profiles & Applications")]
        VectorDB[("Medical Embeddings")]
    end

    %% External APIs
    LLM["External LLM APIs<br>(Groq, Gemini, OpenAI)"]

    %% Connections
    UI <--> API
    API <--> AuthSvc
    API <--> DocSvc
    API <--> SympExt
    
    AuthSvc <--> UsersDB
    DocSvc <--> ProfilesDB
    
    SympExt --> |"Raw Text"| LLM
    LLM --> |"Canonical JSON"| SympExt
    
    SympExt --> |"Canonical Symptoms"| RedFlag
    RedFlag --> |"If safe (No Red-Flags)"| TriageScore
    
    CorpusPipe --> |"Chunking & Embeddings"| VectorDB
    RAG <--> |"Similarity Search"| VectorDB

Loading

Getting Started

This application has been fully containerized using Docker, making it incredibly easy to spin up the Frontend, Backend, and Database with a single command.

Prerequisites

• Docker Desktop installed and running.

Setup Instructions

1. Clone the repository:

   git clone https://github.com/MiliKava/HealthHubAi.git
   cd HealthHubAi

2. Set up Environment Variables: Duplicate the .env.example file inside the backend folder and rename it to .env:

   # On Windows
   copy backend\.env.example backend\.env
   
   # On macOS/Linux
   cp backend/.env.example backend/.env

   (Note: The docker-compose.yml automatically passes the correct DATABASE_URL to the containers, so you do not need to manually configure the database URL unless you are running it outside of Docker).

3. Start the Application: Spin up the entire stack (PostgreSQL with pgvector, FastAPI Backend, and Vite Frontend) in the background:

   docker-compose up -d --build

4. Seed the Database (Optional): To add the default Admin user, run this command inside the running backend container:

   docker-compose exec backend python seed_admin.py

5. Ingest Medical Knowledge (Phase 6): To test the AI Vector Database, you can ingest the MedQuAD dataset into pgvector by running:

   docker-compose exec backend python scripts/ingest_corpus.py

Accessing the Application

• Frontend Interface: Open your browser and navigate to http://localhost:5173.
  • Admin Access: You can log in using admin@healthhub.ai and admin123 to access the Admin Panel to approve doctor applications.
• Backend API Docs: The FastAPI interactive documentation is available at http://localhost:8000/docs.
• Database Access: The PostgreSQL database is mapped to your local port 5433 (to avoid conflicts with local installations). You can connect via pgAdmin using localhost:5433, user postgres, password password, and database healthhub.

---

project is in development phase, stay tuned.