Computational Approaches Toward Integrating Quantified Self Sensing and Social Media

Area of Science:

• Computational social science
• Digital health
• Personalized medicine

Background:

• Quantified self (QS) tools and social media generate vast data for personalized medicine.
• QS data offers physiological signals, while social media provides behavioral insights.
• These data sources have been predominantly analyzed in isolation.

Purpose of the Study:

• To investigate the predictive power of social media data for diet compliance success.
• To explore the relationship between social media features and dietary behavior.
• To assess the potential for integrating social media insights into health interventions.

Main Methods:

• Analysis of public data from users linking MyFitnessPal and Twitter accounts.
• Feature extraction from social media data, including linguistic, activity, and social capital metrics.
• Application of Granger causality methodology to assess predictive relationships.

Main Results:

• Social media features significantly predict diet compliance success.
• Positive affect and larger social networks are associated with higher diet success rates.
• Social media data improved the prediction of daily diet compliance changes by 17% over baseline, achieving 77% accuracy.

Conclusions:

• Social media analytics offer a valuable, previously underutilized, resource for understanding and predicting health behaviors.
• Integrating social media data can enhance the effectiveness of personalized health interventions for behavior change.
• This research highlights the potential of cross-platform data analysis in digital health and personalized medicine.