Understanding the social context of eating is crucial for promoting healthy eating behaviors by providing timely interventions. Multimodal smartphone sensing data has the potential to provide valuable insights into eating behavior, particularly in mobile food diaries and mobile health applications. However, research on the social context of eating with smartphone sensor data is limited, despite extensive study in nutrition and behavioral science. Moreover, the impact of country differences on the social context of eating, as measured by multimodal phone sensor data and self-reports, remains under-explored. To address this research gap, we present a study using a smartphone sensing dataset from eight countries (China, Denmark, India, Italy, Mexico, Mongolia, Paraguay, and the UK). Our study focuses on a set of approximately 24K self-reports on eating events provided by 678 college students to investigate the country diversity that emerges from smartphone sensors during eating events for different social contexts (alone or with others). Our analysis revealed that while some smartphone usage features during eating events were similar across countries, others exhibited unique behaviors in each country. We further studied how user and country-specific factors impact social context inference by developing machine learning models with population-level (non-personalized) and hybrid (partially personalized) experimental setups. We showed that models based on the hybrid approach achieve AUC scores up to 0.75 with XGBoost models. These findings have implications for future research on mobile food diaries and mobile health sensing systems, emphasizing the importance of considering country differences in building and deploying machine learning models to minimize biases and improve generalization across different populations.

翻译：理解饮食的社交情境对于通过提供及时干预来促进健康饮食行为至关重要。多模态智能手机感知数据有望为饮食行为提供有价值的洞察，尤其在移动食物日记和移动健康应用中。尽管营养学和行为科学领域对此已有广泛研究，但基于智能手机传感器数据的饮食社交情境研究仍十分有限。此外，通过多模态手机传感器数据和自我报告测量，国家差异对饮食社交情境的影响仍未被充分探索。为填补这一研究空白，我们开展了一项研究，使用了来自八个国家（中国、丹麦、印度、意大利、墨西哥、蒙古、巴拉圭和英国）的智能手机感知数据集。研究聚焦于678名大学生提供的约24,000份饮食事件自我报告，以探究不同社交情境（独自或与他人共餐）下，智能手机传感器数据所体现的国家多样性。分析表明，尽管各国饮食事件中的部分智能手机使用特征相似，但另一些特征则表现出独特的行为模式。我们进一步通过开发基于群体水平（非个性化）和混合（部分个性化）实验设置的机器学习模型，研究了用户和国家特定因素如何影响社交情境推断。结果显示，采用混合方法的XGBoost模型AUC得分可达0.75。这些发现对未来移动食物日记和移动健康感知系统的研究具有启示意义，强调了在构建和部署机器学习模型时考虑国家差异的重要性，以减少偏差并提高跨人群的泛化能力。