# Understanding the Social Context of Eating with Multimodal Smartphone Sensing: The Role of Country Diversity

**Authors:** Nathan Kammoun, Lakmal Meegahapola, Daniel Gatica-Perez  
**arXiv:** [2306.00709](https://arxiv.org/abs/2306.00709)  
**Date:** 2023-06-01  
**Summary page:** https://www.summarizepaper.com/en/arxiv-id/2306.00709v1/

## Key points

- Social context of eating is crucial in promoting healthy eating behaviors
- Smartphone sensing data can provide valuable insights into eating behavior through mobile food diaries and health applications
- Research on the social context of eating with smartphone sensor data is limited, especially regarding country differences
- A study was conducted using a smartphone sensing dataset from eight countries to investigate the country diversity that emerges from smartphone sensors during eating events for different social contexts (alone or with others)
- Some smartphone usage features during eating events were similar across countries, while others exhibited unique behaviors in each country
- Machine learning models were developed to understand how user and country-specific factors impact social context inference in mobile food diaries and health sensing systems
- The hybrid approach achieved AUC scores up to 0.75 with XGBoost models
- Findings emphasize the importance of considering country differences when building and deploying machine learning models to minimize biases and improve generalization across different populations.

## Summary

The social context of eating plays a crucial role in promoting healthy eating behaviors, and smartphone sensing data can provide valuable insights into eating behavior through mobile food diaries and health applications. However, research on the social context of eating with smartphone sensor data is limited, especially regarding country differences. To address this gap, a study was conducted using a smartphone sensing dataset from eight countries (China, Denmark, India, Italy, Mexico, Mongolia, Paraguay, and the UK) to investigate the country diversity that emerges from smartphone sensors during eating events for different social contexts (alone or with others). The study focused on approximately 24K self-reports on eating events provided by 678 college students. The analysis revealed that while some smartphone usage features during eating events were similar across countries, others exhibited unique behaviors in each country. To further understand how user and country-specific factors impact social context inference in mobile food diaries and health sensing systems, machine learning models were developed with population-level (non-personalized) and hybrid (partially personalized) experimental setups. The hybrid approach achieved AUC scores up to 0.75 with XGBoost models. These findings have implications for future research on mobile food diaries and health sensing systems as they emphasize the importance of considering country differences when building and deploying machine learning models to minimize biases and improve generalization across different populations. Overall, this study highlights the importance of understanding the social context of eating through smartphone sensing data to promote healthy eating behaviors effectively. It also emphasizes the need to consider country differences when developing machine learning models for mobile food diaries and health sensing systems to ensure their effectiveness across diverse populations.