Cognitive Load Typing Tracker

A machine learning-powered web application that analyzes typing patterns in real-time to estimate cognitive load levels. The system tracks various typing metrics (speed, pauses, backspaces, bursts) and uses a trained Gradient Boosting Regressor model to predict mental workload during typing tasks.

Features

1. Real-time Typing Analysis

• Live Typing Test: Type predefined sentences while the app tracks your behavior
• Visual Feedback: Real-time character-by-character feedback (correct/incorrect/current)
• Accuracy Metrics: Track accuracy, correct/incorrect characters, and progress
• Backspace Timeline: Visual timeline showing when corrections were made

2. Cognitive Load Prediction

• ML-Powered Predictions: Uses trained Gradient Boosting model for accurate cognitive load estimation
• Three Load Levels: Low, Medium, High cognitive load classification
• Confidence Score: Behavior-aware confidence metric (40-95%)
• Smoothed Predictions: Rolling buffer to reduce noise and improve stability

3. Mental Health Assessment

• Anxiety Quiz: 10-question assessment with percentage scoring
• Depression Quiz: 10-question PHQ-style assessment
• Stress Quiz: 10-question perceived stress scale
• Interactive Chat Interface: Conversational UI for mental wellness check-ins
• Wellness Dashboard: View all assessment results in one place

4. Dataset Collection & Export

• Label Your Sessions: Mark typing tests as low/medium/high cognitive load
• Session History: Browse and review past typing sessions
• Export Dataset: Download collected data as CSV for model retraining
• Persistent Storage: Local browser storage for history and preferences

Technology Stack

Backend

• Flask: Python web framework for API endpoints
• scikit-learn: Machine learning library (Gradient Boosting Regressor)
• pandas: Data manipulation and feature engineering
• joblib: Model serialization and loading
• Flask-CORS: Cross-origin resource sharing support

Frontend

• Vanilla JavaScript: No framework dependencies
• HTML5/CSS3: Modern, responsive UI with dark/light theme
• LocalStorage: Client-side data persistence
• Fetch API: RESTful API communication

Machine Learning

• Model: GradientBoostingRegressor with 50 estimators
• Features: 10 engineered features (speed, intervals, pauses, corrections, bursts)
• Preprocessing: StandardScaler for feature normalization
• Validation: 5-fold cross-validation for model evaluation

Installation

Prerequisites

• Python 3.7 or higher
• pip package manager
• Modern web browser (Chrome, Firefox, Safari, Edge)

Setup Instructions

1. Clone or extract the project

   cd CTTCProject

2. Install Python dependencies

   pip install -r requirements.txt

   Alternatively, install manually:

   pip install flask flask-cors scikit-learn pandas numpy joblib

3. Verify model files exist The following pickle files should be present:

   • cognitive_load_model.pkl - Trained ML model
   • scaler.pkl - Feature scaler
   • feature_columns.pkl - Feature order reference

4. Start the Flask API server

   python app.py

   The API will start on http://127.0.0.1:5000

5. Open the web interface Open index.html in your web browser (you can double-click the file or use a local server)

Usage Guide

Typing Test

1. Start Test: Open index.html in your browser
2. Type Sentence: Type the displayed sentence in the text area
3. View Real-time Stats: Monitor your typing speed, pauses, backspaces, and cognitive load prediction
4. Complete Test: Click "Complete Test" when finished
5. Label Session: Choose a cognitive load label (low/medium/high) for dataset collection

Mental Health Assessment

1. Access Mental Test: Click "Mental Test" button from main page
2. Choose Assessment: Select Anxiety, Depression, Stress, or "Just fine"
3. Answer Questions: Respond to 10 questions for each assessment type
4. View Results: See your percentage scores and severity levels
5. Wellness Dashboard: Click "View Wellness Dashboard" to see all results

Dataset Management

1. View History: See all past typing sessions in the history panel
2. Review Sessions: Click on history items to replay typing patterns
3. Export Data: Click "Export Dataset as CSV" to download collected data
4. Clear Data: Use "Clear All History" to reset dataset (with confirmation)

Theme Customization

• Click the "Toggle Theme" button to switch between dark and light modes
• Theme preference is automatically saved to localStorage

API Endpoints

POST /predict

Predicts cognitive load based on typing metrics.

Request Body:

{
  "speed": 4.5,
  "avgInterval": 150,
  "pause300": 3,
  "pause500": 1,
  "backspaceCount": 5,
  "maxBurst": 12,
  "editRatio": 0.15,
  "duration": 25.4
}

Response:

{
  "score": 45.23,
  "level": "medium",
  "confidence": 78.5
}

Cognitive Load Levels:

• Low: score < 35 (Minimal mental effort, comfortable typing)
• Medium: 35 ≤ score < 65 (Moderate cognitive demand)
• High: score ≥ 65 (High mental effort, struggling to type)

Model Training

Retrain the Model

To retrain the model with new data:

1. Update dataset.csv with your collected data

   • Required columns: speed, avgInterval, pause300, pause500, backspaceCount, maxBurst, editRatio, duration, label

2. Run training script

   python train_model.py

3. Review metrics

   • Regression: MAE, R² score
   • Classification: Accuracy, F1 score, confusion matrix
   • Cross-validation: 5-fold R² scores

4. Updated models will be saved:

   • cognitive_load_model.pkl
   • scaler.pkl
   • feature_columns.pkl

Feature Engineering

The model uses 10 engineered features:

Feature	Description
speed	Characters per second
avgInterval	Average time between keystrokes (ms)
pause300	Number of pauses > 300ms
pause500	Number of pauses > 500ms
backspaceCount	Total backspace/delete presses
maxBurst	Longest continuous typing burst
editRatio	Ratio of edits to total characters
duration	Total typing duration (seconds)
pause_ratio_300	pause300 / total_keys
correction_rate	backspaceCount / total_keys

Project Structure

CTTCProject/
├── app.py                     # Flask API server
├── train_model.py             # Model training script
├── dataset.csv                # Training dataset
├── cognitive_load_model.pkl   # Trained ML model
├── scaler.pkl                 # Feature scaler
├── feature_columns.pkl        # Feature order reference
├── index.html                 # Main typing tracker UI
├── mental.html                # Mental health
├── requirements.txt           # Requirement file
├── wellness.html              # Wellness dashboard UI
└── README.md                  # This file

Keyboard Shortcuts

Main Typing Interface

• Enter: Submit typing input
• Esc: Reset test
• ?: Show keyboard shortcuts help

Browser Compatibility

• Chrome 90+ (recommended)
• Firefox 88+
• Safari 14+
• Edge 90+

Data Privacy

• All data is stored locally in your browser's localStorage
• No data is sent to external servers except the local Flask API
• Mental health assessments are completely private
• Export/delete your data at any time

Troubleshooting

API Connection Issues

• Ensure Flask server is running on port 5000
• Check console for CORS errors
• Verify flask-cors is installed

Model Not Loading

• Confirm all .pkl files exist in the project directory
• Check Python version compatibility (3.7+)
• Verify scikit-learn version matches training environment

Predictions Seem Inaccurate

• Collect more labeled data and retrain
• Ensure consistent typing behavior during labeling
• Check feature engineering matches training

LocalStorage Full

• Export dataset and clear history
• Browser storage limits vary (typically 5-10MB)

Future Enhancements

• Real-time stress monitoring with webcam (facial expressions)
• Multi-language support for typing tests
• Cloud storage integration for cross-device sync
• Advanced analytics dashboard with charts
• Mobile app version
• Calibration mode for personalized predictions
• Integration with productivity tools

Contributing

Contributions are welcome! Areas for improvement:

• Additional typing test sentences and difficulty levels
• Enhanced mental health assessment questions
• Improved ML models (LSTM, Transformer architectures)
• UI/UX enhancements
• Accessibility improvements

License

This project is open source and available for educational and research purposes.

Acknowledgments

• Typing pattern analysis inspired by cognitive psychology research
• Mental health assessments based on GAD-7, PHQ-9, and PSS-10 scales
• Machine learning approach adapted from keystroke dynamics literature

Contact

For questions, issues, or suggestions, please create an issue in the project repository.

---

Note: This tool is for research and educational purposes only. Mental health assessments are not a substitute for professional psychological evaluation or treatment. If you're experiencing mental health concerns, please consult a qualified healthcare professional.