MindPulse: An AI-Powered Mental Wellness Prediction Platform Using Ensemble Machine Learning

MindPulse is a privacy-first, locally-operated, AI-powered mental wellness prediction platform built with
Python (Flask), Scikit-learn, and HTML/CSS/JavaScript. The system leverages a Voting Ensemble model
combining Gradient Boosting and Random Forest classifiers to categorize users' mental states across five
wellness levels: Excellent, Good, Moderate Concern, High Concern, and Critical. The model is trained on
3,000 synthetically generated samples calibrated to established clinical instruments—the PHQ-9, GAD-7, and
WHO-5 Wellbeing Index. Users rate ten behavioral and emotional wellness indicators on a 1–10 scale through
an animated web interface, receiving a composite Wellness Score (0–100), Canvas-rendered Radar Chart,
probability breakdown, and personalized therapeutic insights. Operating entirely on a local Flask server
without any external API keys or cloud transmission ensures complete data privacy. Rigorous testing
comprising 29 test cases across unit, integration, boundary value, and user acceptance testing yielded a 100%
pass rate.