🩺 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 with precautions, OTC suggestions, and doctor alerts for critical cases.

⚠️ 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, temp, and free-text symptoms.
• Real-time Inference: Flask loads model.pkl and predicts on submit.
• Recommendations: OTC guidance, home remedies, doctor alerts.
• Bilingual UX: English + Urdu forms, labels, and results.
• History & Logs: View past predictions per user.
• Admin Dashboard: Users overview, audit, model health (basic).
• MySQL-backed: Persistent storage for users and history.

---

🧩 Architecture

flowchart TD
    A[User (Web/Mobile)] -->|Forms/Requests| B[Flask App]
    B --> C[Validation + Preprocess]
    C --> D[ML Model (model.pkl)]
    D --> E[Prediction + Confidence]
    E --> F[Recommendation Engine]
    F --> G[Response Formatter (EN/UR)]
    B <-->|ORM/SQL| H[(MySQL)]

Loading

---

🛠 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
FLASK_ENV=development
SECRET_KEY=change_this_secret

# DB
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}

# ML
MODEL_PATH=ml/model.pkl

# i18n
DEFAULT_LANG=en
FALLBACK_LANG=en

5) DB Init (example)

python manage.py db upgrade   # or run provided init script

6) Run

flask --app app run --debug
# http://127.0.0.1:5000

---

🔧 Configuration

• Language: Default via DEFAULT_LANG. Toggle per-user in profile.
• Model: Path configurable via MODEL_PATH. Replace ml/model.pkl with a new export when retraining.
• 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 & symptoms → prediction
GET	/history	Current user’s prediction history
POST	/logout	End session

/predict — Request

{
  "height_cm": 175,
  "weight_kg": 78,
  "bp_systolic": 120,
  "bp_diastolic": 80,
  "temperature_c": 37.2,
  "symptoms": "headache, fatigue, mild cough"
}

/predict — Response

{
  "prediction": "Viral Infection (mild)",
  "confidence": 0.82,
  "advice": {
    "medicines": ["paracetamol 500mg (if needed)", "oral hydration salts"],
    "home_remedies": ["rest", "warm fluids"],
    "exercises": ["light stretching only"],
    "consult_doctor": false,
    "alerts": ["if fever > 38.5°C for 48h, seek medical attention"]
  },
  "lang": "en"
}

---

🗂 Project Structure

Atria-AI/
├── app/
│   ├── __init__.py
│   ├── routes.py
│   ├── models.py
│   ├── services/
│   │   ├── inference.py        # loads model.pkl, schema, predict()
│   │   └── recommend.py        # rules for meds/exercises/alerts
│   ├── i18n/
│   │   ├── en.json
│   │   └── ur.json
│   ├── templates/
│   │   ├── base.html
│   │   ├── auth/
│   │   │   ├── login.html
│   │   │   └── register.html
│   │   └── dashboard/
│   │       ├── patient.html
│   │       └── admin.html
│   └── static/
│       ├── css/
│       └── js/
├── 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
├── ml/
│   ├── data/
│   ├── notebooks/
│   ├── training.py
│   └── model.pkl
├── migrations/
├── tests/
├── requirements.txt
├── .env.example
├── manage.py
└── README.md

---

🔒 Security & Privacy

• Passwords: Hash + salt (e.g., werkzeug.security / passlib).
• CSRF: Protect forms (e.g., Flask-WTF).
• Validation: Strict input schema (e.g., pydantic/marshmallow).
• Logs: Avoid PHI; redact sensitive fields.
• RBAC: Reserved for admin features and future doctor roles.
• Headers: Enforce HTTPS, HSTS, CSP, and secure cookies in production.

---

🗺 Roadmap

• Symptom NLP (entities + negation handling)
• Differential multi-condition outputs
• Doctor directory + appointment integration
• Model monitoring + drift detection
• Dockerfile + CI/CD (GitHub Actions)
• PWA features + 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 or before taking any medication.

---

🗣 Urdu (اردو) Summary

AtriaAI ایک ویب بیسڈ سسٹم ہے جو علامات اور بنیادی ویٹلز (قد، وزن، بلڈ پریشر، درجہ حرارت) کی بنیاد پر ابتدائی خود جانچ میں مدد دیتا ہے۔

• ممکنہ بیماری/حالت کی پیش گوئی
• احتیاطی تجاویز، گھریلو علاج، ہلکی ورزشیں
• ڈاکٹر سے رجوع کی وارننگ اگر مسئلہ سنگین ہو

یہ سسٹم ڈاکٹر کا متبادل نہیں، صرف ابتدائی رہنمائی کے لیے ہے۔

---

🗂 Assets Inventory

Place these files in the repo to match references above:

assets/
├── banner/
│   └── atriavai-banner.svg                # top README banner
└── screens/
    ├── login-light.png                    # login page (light)
    ├── login-dark.png                     # login page (dark)
    ├── dashboard-patient-light.png        # patient dashboard (light)
    ├── dashboard-patient-dark.png         # patient dashboard (dark)
    ├── dashboard-admin-light.png          # admin dashboard (light)
    └── dashboard-admin-dark.png           # admin dashboard (dark)