🧠 HealthAI

An AI-powered diagnostic assistant designed to enhance rural healthcare by addressing language barriers, inconsistent records, and limited diagnostic tools.

---

🚑 Problem Statement

Rural healthcare in India faces significant challenges:

• Language Barriers: Diverse languages hinder accurate assessments.
• Inconsistent Records: Manual paperwork leads to incomplete patient histories.
• Limited Diagnostic Tools: Lack of advanced equipment and specialists delays diagnoses.

---

💡Solution

An AI-driven platform that assists doctors through:

1. Conversation-to-Assessment Assistant:

   • Records doctor-patient interactions in any language.
   • Translates speech to English using advanced NLP models.
   • Extracts symptoms, suggests diagnoses, and highlights red flags via LLMs.
   • Presents a concise, visual dashboard of the case assessment.

2. Diagnostic Interpretation Assistant:

   • Interprets data from diagnostic devices (e.g., ECGs, X-rays).
   • Uses pre-trained models to identify abnormalities.
   • Empowers general physicians with informed decision-making tools.

---

💰 Business Model

1. B2B SaaS: Offer subscription-based access to hospitals and clinics, providing scalable AI diagnostics and multilingual assistance.

2. Freemium Model: Provide essential features for free to encourage adoption, with premium diagnostics and advanced analytics available through paid tiers.

3. API Licensing: License our AI diagnostic APIs to medical device manufacturers, enabling integration into their products and expanding our reach.

4. Data Analytics Services: Offer anonymized health data insights to researchers and public health agencies, ensuring compliance with data privacy regulations.

5. Training & Certification: Provide online training modules and certification programs for healthcare professionals to ensure effective use of our platform.

---

📁 File Structure

HEALTHAI/
├── backend/
│ ├── fastapi/ # Python backend for clinical NLP and ML models
│ └── node-backend/ # Node.js backend for user authentication and session management
├── ecgbackend/ # Backend for ECG data processing
├── frontend/ # React-based responsive frontend
├── .gitignore
├── README.md

---

🚀 How to Run the Project

🧩 Prerequisites

Ensure you have the following installed on your system:

• Python: Version 3.9 or higher

• Node.js: Version 18 or higher

• npm or yarn

• virtualenv (for Python environments)

• Git (to clone the repository)

• GPU (recommended for running models like Whisper and LLMs)

  📥 Step 1: Clone the Repository

Open your terminal and execute:

git clone https://github.com/Nehilsa2/HealthAI.git
cd HEALTHAI

🔧 Step 2: Set Up the Backend

a. FastAPI (Python Backend)

cd backend/fastapi
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate
pip install -r requirements.txt
uvicorn main:app --reload

This will start the FastAPI server on http://127.0.0.1:8000.

b. Node.js Backend

cd ../node-backend
npm install
node index.js

This will start the Node.js server on the default port (usually http://localhost:3001).

💻 Step 3: Set Up the Frontend

cd ../../frontend
npm install
npm start

This will start the React frontend on http://localhost:3000.

✅ Step 4: Access the Application

Open your browser and navigate to http://localhost:3000 to interact with the HealthAI application.

---

🛠️ Tech Stack

Layer	Tools Used
Frontend	React.js, Tailwind CSS
Backend	FastAPI (Python), Node.js
Speech & NLP	Whisper, Google Speech-to-Text, NLLB
LLM Analysis	GPT-4, LLaMA (fine-tuned on medical data)
Diagnostics	CNN Models (ECG, X-rays e.g., CheXNet)

---

🧬 Model Architecture Overview

Our system integrates multiple AI models to enhance diagnostic accuracy and efficiency:

1. Speech Recognition & Translation

• Whisper: Transcribes multilingual doctor-patient conversations into text.
• NLLB (No Language Left Behind): Translates transcriptions into English, ensuring language inclusivity.

2. Clinical NLP & Assessment

• GPT-4 / LLaMA (fine-tuned): Processes translated conversations to extract symptoms, suggest possible diagnoses, and highlight critical red flags.

3. Diagnostic Image & Signal Interpretation

• CheXNet: Analyzes chest X-rays to detect potential abnormalities.
• Custom CNN Models: Processes ECG data to identify irregularities and assist in cardiac assessments.

These models collectively provide a comprehensive diagnostic overview, aiding physicians in making informed decisions.

---

📊 Impact and Use Cases

• General Physicians in under-resourced areas.
• Mobile Health Clinics and NGO healthcare providers.
• Public Health Missions like Ayushman Bharat and NHM.

---

📎 References

• ChatGPT for Discharge Summaries - J Korean Med Sci. 2024
• MedRxiv: Bridging Rural Health Gap