Explore the intersection of machine learning and mental health in this 22-minute conference talk from ShmooCon 2019. Delve into the capabilities of ML models to predict mental health status through Twitter data analysis, and examine the associated privacy implications. Discover how these models can identify individuals with mental illnesses even without explicit mentions in their social media posts. Analyze various linguistic features, including bag of words, word clusters, and topic models, to understand the underlying patterns differentiating individuals with mental health conditions from control groups. Consider potential applications, feasibility, and privacy concerns surrounding this technology. Learn about mitigating steps for policymakers, social media platforms, and users to address these challenges. Gain insights from Janith Weerasinghe, a doctoral candidate specializing in machine learning and privacy, and Dr. Rachel Greenstadt, an Associate Professor of Computer Science and Engineering with extensive experience in hacker conferences.
Overview
Syllabus
Introduction
Feature Analysis
Applications
Misclassification
Mitigation
Conclusion
Taught by
0xdade