Скачать или смотреть Early Mental Health Risk Detection with Machine Learning

This video shows how we use large-scale survey data to screen for early mental-health risk in schools and workplaces. We analyze ~300k anonymous responses with 17 features covering demographics, stress levels, work context, and coping indicators (Jikadara, 2024). The goal is a trustworthy, data-driven risk screener that flags low, moderate, or high risk so support reaches people sooner. The tool augments counselors and wellness teams and does not replace clinical judgment. The public health need is significant, with depression affecting 280+ million people worldwide (World Health Organization, 2023).

Method overview: exploratory analysis, preprocessing, and supervised modeling with Logistic Regression, a Tabular Neural Network inspired by TabNet, and a soft-voting ensemble. We emphasize calibration, fairness checks, privacy controls on anonymous data, and practical deployment via a simple web interface.

Project links and credits:
GitHub repo: https://github.com/oxayavongsa/aai-59...
Dataset: https://www.kaggle.com/datasets/bhavi...
Course: University of San Diego, Shiley-Marcos School of Engineering, AAI-590 Capstone
Team: Outhai Xayavongsa (Team Lead), Aaron Ramirez (Tech Lead), Prema Mallikarjunan
Advisor: Professor Anna Marbut

#mentalhealth #MachineLearning #RiskScreening #Wellness #Education #Workplace