About Our Research

Academic Research Initiative

This application demonstrates a sentiment-based machine learning approach to estimating the likelihood of PTSD from text transcripts. It is designed for research and educational purposes only and is not intended as a medical diagnostic tool.

100% Privacy Protected - All Processing Local
Sentiment Analysis
  • VADER: Lexicon-based sentiment analyzer
  • TextBlob: Pattern-based polarity detection
  • Real-time emotion and sentiment scoring
Feature Engineering
  • 23-bin sentiment discretization
  • Statistical aggregations and proportions
  • Text length and structure analysis
Ensemble Learning
  • Random Forest, Gradient Boosting
  • Linear SVC, Linear Discriminant Analysis
  • Super Learner stacking methodology

Important Limitations

Language & Context
  • English language only
  • Optimized for interview-style transcripts
  • No cultural or demographic adjustments
Data & Validation
  • Trained on synthetic demo data
  • No clinical validation studies
  • Limited dataset size constraints
Temporal Analysis
  • No longitudinal modeling
  • Single-point-in-time analysis
  • No progression tracking
Clinical Scope
  • Not a diagnostic instrument
  • Requires professional interpretation
  • Educational demonstration only
Research Applications
  • Machine learning education in healthcare
  • Sentiment analysis methodology demonstration
  • Ensemble learning technique illustration
  • Natural language processing research
  • Mental health informatics training
  • Feature engineering case studies

Built for advancing healthcare AI research and education

Try the Interactive Demo