Suicide Risk Classification Web App

An AI-powered text analysis tool with explainable predictions using SHAP and LIME for suicide risk detection. See report at report.pdf.

Model Information

• Accuracy: 92%
• Base Model: BERT (bert-base-uncased)
• Dataset: Kaggle Suicide and Depression Detection (12,000+ samples)
• Labels: Binary classification (Suicidal / Non-Suicidal)

Features

• Real-time text classification
• SHAP (SHapley Additive exPlanations) visualizations
• LIME (Local Interpretable Model-agnostic Explanations) analysis
• Minimalist black & white UI
• REST API endpoint
• Production-ready with Gunicorn

Prerequisites

• Python 3.10+
• Git
• Git LFS (for model files)
• UV package manager (recommended) or pip

Quick Setup

Option 1: Using UV (Recommended)

# Clone repository
git clone git@github.com:creme332/suicide-detector.git
cd suicide-detector

# Create virtual environment and install dependencies
uv venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate
uv pip install -r requirements.txt

# Run the app
uv run python app.py

Option 2: Using pip

# Clone repository
git clone git@github.com:creme332/suicide-detector.git
cd suicide-detector

# Create virtual environment
python -m venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

# Run the app
python app.py

Access the app at http://localhost:5000 in your browser.

API Usage

curl -X POST http://localhost:5000/api/predict \
  -H "Content-Type: application/json" \
  -d '{"text": "I feel hopeless and dont want to live"}'

Tech Stack

• Backend: Flask
• ML Framework: PyTorch, Transformers
• Explainability: SHAP, LIME
• Frontend: HTML, CSS, JavaScript
• Deployment: Gunicorn

Collaborators

• creme332
• Divyeshhhh