🩺 AtriaAI — Intelligent Early Medical Diagnosis System

Web-based ML-powered assistant for early symptom assessment with bilingual (English/Urdu) support.

---

📑 Table of Contents

• Overview
• Objectives
• Key Features
• Architecture
• Tech Stack & Badges
• Screenshots
• Quickstart
• Configuration
• API Endpoints
• Project Structure
• Security & Privacy
• Roadmap
• Contributing
• License
• Disclaimer
• Urdu (اردو) Summary
• Assets Inventory

---

🧠 Overview

AtriaAI helps patients perform an early self-assessment using symptoms and vitals (height, weight, BP, temperature). A trained ML model (model.pkl) runs on the server and returns a possible condition, confidence, and practical recommendations.

⚠️ AtriaAI does not replace clinical diagnosis. It is an assistive tool for awareness and timely consultation.

---

🎯 Objectives

• Enable quick, simple self-checks via a clean, accessible UI.
• Provide bilingual (English/Urdu) outputs with non-technical explanations.
• Suggest precautions, home remedies, and exercise tips.
• Trigger consultation alerts for red-flag scenarios.

---

✅ Key Features

• Auth & Profiles: Registration, login, profile management.
• Vitals & Symptoms: Height, weight, BP, temperature, and free-text symptoms.
• Real-time Inference: Flask loads model.pkl and predicts on submit.
• Recommendations: OTC guidance, home remedies, and doctor alerts.
• Bilingual UX: English and Urdu forms, labels, and results.
• History & Logs: View past predictions per user.
• Admin Dashboard: Users overview, audit, and basic model health.
• MySQL-backed: Persistent storage for users and history.

---

🧩 Architecture

flowchart TD
    U[User Web or Mobile Client] -->|Forms and API requests| W[Flask Web Application]
    W --> AUTH[Authentication and Profiles]
    W --> V[Vitals and Symptom Input]
    V --> VAL[Validation and Preprocessing]
    VAL --> M[ML Inference Service<br/>model.pkl]
    M --> P[Prediction and Confidence]
    P --> R[Recommendation Engine]
    R --> F[Response Formatter<br/>English or Urdu]
    F --> UI[Dashboard and Prediction Result]

    W --> H[History and Audit Services]
    H --> DB[(MySQL Database)]
    AUTH --> DB
    H --> ADM[Admin Dashboard]
    ADM --> DB

    W --> SEC[Security Controls<br/>CSRF, RBAC, HTTPS, Secure Cookies]

Loading

The diagram reflects the documented Flask application, authentication and profiles, validated vitals/symptoms flow, model inference, bilingual response formatting, history, admin dashboard, MySQL persistence, and production security controls.

---

🛠 Tech Stack & Badges

---

🖼 Screenshots

---

🚀 Quickstart

1) Clone

git clone https://github.com/UsamaMatrix/Atria-AI.git
cd Atria-AI

2) Virtual Env

python -m venv .venv
# Windows
.venv\Scripts\activate
# macOS/Linux
source .venv/bin/activate

3) Install

pip install -r requirements.txt

4) Configure .env

FLASK_ENV=development
SECRET_KEY=change_this_secret
DB_HOST=127.0.0.1
DB_PORT=3306
DB_USER=atria_user
DB_PASSWORD=strong_password
DB_NAME=atria_db
DATABASE_URL=mysql+pymysql://${DB_USER}:${DB_PASSWORD}@${DB_HOST}:${DB_PORT}/${DB_NAME}
MODEL_PATH=ml/model.pkl
DEFAULT_LANG=en
FALLBACK_LANG=en

5) DB Init (example)

python manage.py db upgrade

6) Run

flask --app app run --debug

---

🔧 Configuration

• Language: Default via DEFAULT_LANG; toggle per user in the profile.
• Model: Configure the path with MODEL_PATH.
• Security: Use a strong SECRET_KEY; enable HTTPS and secure cookies in production.

---

🔌 API Endpoints

Method	Endpoint	Description
GET	/	Health check / landing
GET	/login	Login form
POST	/login	Authenticate
GET	/register	Registration form
POST	/register	Create new user
GET	/dashboard	User dashboard
POST	/predict	Submit vitals and symptoms
GET	/history	Current user's prediction history
POST	/logout	End session

---

🗂 Project Structure

Atria-AI/
├── app/
│   ├── __init__.py
│   ├── routes.py
│   ├── models.py
│   ├── services/
│   │   ├── inference.py
│   │   └── recommend.py
│   ├── i18n/
│   │   ├── en.json
│   │   └── ur.json
│   ├── templates/
│   └── static/
├── assets/
├── ml/
│   ├── data/
│   ├── notebooks/
│   ├── training.py
│   └── model.pkl
├── migrations/
├── tests/
├── requirements.txt
├── .env.example
├── manage.py
└── README.md

---

🔒 Security & Privacy

• Passwords: Hash and salt passwords.
• CSRF: Protect forms with CSRF tokens.
• Validation: Apply strict input validation.
• Logs: Avoid PHI and redact sensitive fields.
• RBAC: Restrict admin features.
• Headers: Enforce HTTPS, HSTS, CSP, and secure cookies in production.

---

🗺 Roadmap

• Symptom NLP (entities and negation handling)
• Differential multi-condition outputs
• Doctor directory and appointment integration
• Model monitoring and drift detection
• Dockerfile and CI/CD
• PWA features and offline cache

---

🤝 Contributing

1. Fork the repo.
2. Create a feature branch.
3. Commit with clear messages.
4. Open a PR with context and screenshots.

📄 License

Licensed under the MIT License. See LICENSE.

⚠️ Disclaimer

AtriaAI provides informational guidance only and does not replace professional medical diagnosis, treatment, or advice. Always consult a qualified healthcare provider for serious or persistent symptoms.

🗣 Urdu (اردو) Summary

AtriaAI ایک ویب بیسڈ سسٹم ہے جو علامات اور بنیادی ویٹلز کی بنیاد پر ابتدائی رہنمائی فراہم کرتا ہے۔ یہ ڈاکٹر کا متبادل نہیں ہے۔

🗂 Assets Inventory

assets/
├── banner/
│   └── atriavai-banner.svg
└── screens/
    ├── login-light.png
    ├── login-dark.png
    ├── dashboard-patient-light.png
    ├── dashboard-patient-dark.png
    ├── dashboard-admin-light.png
    └── dashboard-admin-dark.png