CAREER: Interactive Dietary Monitoring

虽然现在使用智能手表和其他可穿戴设备跟踪睡眠和步行等身体活动已经很普遍，但跟踪一个人的饮食仍然是一项繁重的任务。从健康的角度来看，这是一个错失的机会；饮食自我监控是人们反思和改善饮食习惯的有效途径，可以减少患肥胖等与饮食有关的慢性疾病的机会。该项目通过提出一种称为交互式饮食监测的新方法，解决了现有饮食监测方法的弱点。这个想法是将饮食跟踪设想为人机协作，结合手动和自动食物跟踪的优势。例如，人们知道他们正在吃什么食物，但可能会忘记在那一刻做笔记。自动化系统可以检测到人们什么时候在吃东西，但对食物身份的了解有限。因此，协作方法可以弥补互补的弱点。为了将这一愿景变为现实，项目团队将原型化、评估和改进用于食物跟踪的新型可穿戴设备、算法和视觉界面，目的是减轻饮食自我监控的负担，同时让个人了解自己的饮食行为。开发的设备和方法可能对其他紧迫问题有用，包括标记与饮食失调和吸烟事件相关的行为，测量帕金森病患者的震颤，并帮助营养流行病学家了解饮食和疾病之间的联系。该项目将通过(1)开发和评估检测细粒度手指运动的算法和围绕饮食跟踪的新型交互技术，(2)研究酒精饮酒检测和从人类呼吸中存在的挥发性有机化合物中识别食物的传感方法，(3)产生关于开发用于饮食跟踪的可穿戴设备的新知识，推进人机交互和泛在计算领域的发展。(4)探索一种新的移动食品杂志界面，利用颜色和营养之间的关系来提供视觉饮食见解；(5)通过纳入老年人、慢性病老年人和研究中代表性不足的个体，产生可用性和算法发现。该提案的成果将通过以人为中心的过程进行开发和评估，该过程包括在受控和自然条件下的形成性研究、实施和评估。为了增强本研究的影响力和社会影响力，项目成果将嵌入研究生和本科生课程，可穿戴平台将整合到经济困难学生的科学营中，高中实习生将在研究的执行中发挥作用。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

While tracking sleep and physical activity such as walking is now commonplace using smartwatches and other wearable devices, keeping track of what a person eats and drinks remains an onerous task. From a health perspective, this is a missed opportunity; dietary self-monitoring is an effective way for people to reflect on and improve their eating habits, reducing the chance of developing costly diet-related chronic diseases such as obesity. This project addresses the weaknesses of existing dietary monitoring methods by advancing a new approach called Interactive Dietary Monitoring. The idea is to envision dietary tracking as a human-machine collaboration, combining the strengths of manual and automated food tracking. For example, people know what food they are eating, but can forget to make a note of it in the moment. Automated systems can detect when people are eating, but have limited insight into the identity of the foods. Thus, a collaborative approach can shore up the complementary weaknesses. To move this vision toward reality, the project team will prototype, evaluate, and refine novel wearable devices, algorithms, and visual interfaces for food tracking, with the goal of reducing the burden of dietary self-monitoring while keeping individuals aware of their eating behaviors. The devices and methods developed are likely to be useful for other pressing problems, including flagging behaviors associated with eating disorders and smoking events, measuring tremor in Parkinson’s disease patients, and helping nutritional epidemiologists understand the link between diet and disease.This project will advance the fields of human-computer interaction and ubiquitous computing through (1) developing and evaluating algorithms to detect fine-grained finger motion and novel interaction techniques around diet tracking, (2) investigating sensing methods for alcohol drinking detection and food recognition from volatile organic compounds present in human breath, (3) producing new knowledge about developing wearable devices for dietary tracking, (4) exploring a new mobile food journal interface that leverages the relationship between color and nutrients to provide visual dietary insights, and (5) generating usability and algorithmic findings from the inclusion of older adults, seniors with chronic diseases, and under-represented individuals in studies. The outcomes of this proposal will be developed and assessed through a human-centered process consisting of formative studies, implementations, and evaluations in both controlled and naturalistic conditions. To enhance the impact and social reach of this research, the outcomes of the project will be embedded into graduate and undergraduate courses, the wearable platforms will be integrated into a science camp for economically disadvantaged students, and high school interns will play a role in the execution of the research.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.