amirsoahil101/MindPulse_AI_Student_Mental_Health_Prediction_Model

🧠 MindPulse AI — Student Mental Health Prediction Web App

MindPulse AI is a Machine Learning-powered full-stack web application that predicts a student's Mental Health Score based on their daily habits (sleep duration, study hours, social media usage, physical activity, etc.).

---

🌐 Live Demo & UI Preview

• Live Application: Click Here to Access MindPulse AI

📸 Application Interface

(Replace this placeholder with your actual UI screenshot path or link)

---

🚀 Old Flow vs. New Flow (Project Evolution)

This project represents a major architectural upgrade in my Machine Learning engineering workflow towards production-ready standards:

Feature / Step	🔴 Old Flow	🟢 New Flow (Upgraded)
Deployment	Streamlit UI	Render Web Service (FastAPI Backend + HTML/JS Frontend)
Preprocessing	Manual Ad-hoc Transformations	Scikit-learn ColumnTransformer (Automated & Leak-free)
Model Training	Manual Loops & Ad-hoc Scripts	Integrated ML Pipelines (Pipeline([('prep', ...), ('model', ...)]))
API & Validation	Raw JSON / Basic Flask	FastAPI + Pydantic Schema Validation
Frontend	Default Streamlit UI	Custom UI (Vanilla HTML5, Modern CSS, JavaScript Fetch API)

---

🔮 Future Integration Plan (Upgrading Legacy Projects)

Through this project, I have mastered end-to-end production ML pipelines and modular deployment strategies. Going forward, I will retrofit and upgrade all my legacy ML and Deep Learning (DL) projects with these same features:

• ⚙️ End-to-End Pipelines: Transitioning manual preprocessing and model fitting into unified Scikit-learn pipelines.
• ⚡ FastAPI Migration: Replacing legacy Streamlit / Flask endpoints with fast, asynchronous, and modular FastAPI backends.
• 🛡️ Pydantic Validation & Security: Implementing strict schema validation, robust error handling, and environment-driven configurations (.env) to securely manage API keys and backend endpoints.

---

🛠️ Project Architecture & Tech Stack

1. Machine Learning Pipeline (notebook.ipynb):

   • Data cleaning, log transformations, and outlier treatment.
   • Categorical feature encoding and scaling via ColumnTransformer.
   • Model selection & evaluation: Linear Regression vs. Random Forest (Default & Tuned).
   • Pipeline serialization using joblib.

2. Backend API (main.py):

   • FastAPI: Asynchronous and high-performance Web API framework.
   • Pydantic: Input schema validation with field-level constraints (age, study_hours, stress_level, etc.).
   • CORS Middleware: Safe cross-origin request handling.

3. Frontend (index.html, script.js, style.css):

   • Interactive gauge dashboard for visual score representation.
   • Dynamic error handling for both client-side and server-side validation.

---

📊 Model Performance Metrics

Model	Test $R^2$	Training $R^2$	MAE	RMSE
Linear Regression	0.7397	0.7236	0.5361	0.6760
Random Forest (Default)	0.8775	0.9808	0.3472	0.4636
Random Forest (Tuned)	0.8650	0.9547	0.3689	0.4869

---

📂 Repository Directory Structure

.
├── Mental_Health_Model.pkl           # Trained Keras Artificial Neural Network (ANN) model
├── notebook.ipynb                    # Jupyter Notebook (EDA, Preprocessing, ANN Training)
├── Student_Social_Media              # Dataset containing audio/speech acoustic features
├── main.py                           # FastAPI Backend
├── UI.png                            # UI Screenshot
├── .gitignore                        # Git ignore file
├── .gitattributes                    # Git attributes file
├── requirements.txt                  # Python dependencies
├── style.css                         # CSS for Streamlit frontend
├── script.js                         # JavaScript for Streamlit frontend
└── index.html                        # Streamlit frontend

git clone https://github.com/amirsohail100/MindPulse-AI-Student-Mental-Health-Prediction-Web-App.git

cd MindPulse-AI-Student-Mental-Health-Prediction-Web-App

streamlit run app.py

pip install -r requirements.txt

📄 License

This project is licensed under the MIT License.

📝 Author

👤 Amir Sohail

MindPulse AI: A production-grade ML web app predicting student mental health scores based on daily lifestyle habits. Features an upgraded workflow: Scikit-learn Pipeline & ColumnTransformer for leak-free training, FastAPI + Pydantic for schema validation, and an interactive HTML/JS frontend live deployed on Render.