Social Media Image Analysis for Public Health
summary
200,000 Instagram food pictures from the 100 most populous counties form the dataset and a supervised regression uses county demographics plus user-provided tags and Imagga machine-generated tags to predict county-level health stats (obesity, diabetes, excessive drinking). Including machine tags improves models over demographics alone, and for excessive drinking tags like liquid and glass from the vision system outperform user tags, showing visual descriptors add measurable signal for population health prediction.
AI-written summary · about
chapters (6)
- 0:00Start
- 1:24Why Social Media?
- 2:14So Far: Using Social Media Text to Model Health Statistics
- 6:53Dependent Variables: Health Statistics
- 7:56Excessive Drinking: What Images Reveal
- 8:44Questions?
related talks
18mInstagram in focus
39mImpacting Global Policy by Understanding Litter Data
18mMultimodal Classification of Moderated Online Pro-Eating Disorder Content
06mOne More Bite? Inferring Food Consumption Level of College Students Using Smartphone Sensing and ...
41mOCR in the Wild World of Social Media
42mModeling Aesthetic Identity- Building a Digital Twin
01h 03mUsing Machine Learning and AI to serve people and communities
06mExamining the Social Context of Alcohol Drinking in Young Adults with Smartphone Sensing