EAGER: Personal Models for a Navigational Approach to Health

上个世纪的科学和技术进步帮助人类摆脱了传染病作为主要死亡原因的时代。发达国家的平均寿命已从40岁增加到80岁以上。今天，慢性病是我们最大的健康挑战。目前基于偶发性传染病健康模式的卫生保健实践并不是治疗慢性疾病的好方法。相反，医疗保健需要关注饮食、锻炼和其他生活方式活动。未来的卫生系统将使用丰富的多模式数据来持续感知健康状态，并为日常生活方式决策提供指导。生活方式导航方法是一个基于传感器、计算元素和执行器的闭环感知-估计-行动循环，并密切受到个人模型的指导。为了提供指导、预测未来或了解个人偏好，我们必须建立一个针对个人的模型。个人模型不仅包含普通的医疗记录数据，还包含疾病倾向和生活方式行为的信息。该项目将开发方法，方便、准确地跟踪对健康和慢性疾病有影响的行为，并将这些信息纳入个性化的生活方式健康模型。这些模型对于制定健康导航性方法，以促进基于预测和预防方法的精准医学至关重要。营养是生活方式的主要组成部分之一，但目前还没有完全令人满意的方法来记录食物。该提案将特别创建一种新的食物记录方法，该方法基于使用可穿戴传感器来检测与进食相关的心率变化，从而触发食物记录的生态瞬间评估（EMA）。与已有的活动日志相比，食物日志提供了足够的定量信息，可以定量地了解和分析一个人的生活方式。该项目将通过扩展机器学习方法来开发新的事件挖掘技术，为一个人建立与生活方式相关的因果模型，该模型将用于预测方法以及指导人们的生活方式决策。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。

Scientific and technological advances in the last century have helped humankind move past infectious disease as the primary cause of mortality. Average lifespan has doubled from 40 years to above 80 years in developed countries. Today, chronic disease presents our greatest health challenge. Current healthcare practice based on the episodic infectious disease health model is not a good approach to treating chronic disease. Instead, healthcare needs to focus on diet, exercise, and other lifestyle activities. Future health systems will use rich, multi-modal data to sense health state continuously and provide guidance in making daily lifestyle decisions. The lifestyle navigational approach is a close-loop sense-estimate-act cycle based on sensors, computing elements, and actuators and closely guided by the personal model of the individual. To provide guidance, predict the future, or understand preferences of an individual, we must build a model that is specific to an individual. The personal model contains not only ordinary medical record data, but also information about disease propensity and lifestyle behaviors.This project will develop methods to easily and accurately track behaviors that have an impact on health and chronic disease, and methods for incorporating such information into a personalized lifestyle health model. These models will be critical in developing a navigational approach to health in order to facilitate precision medicine based on predictive and preventive approaches. Nutrition is one of the major components of lifestyle, and yet there are no fully satisfactory approaches for food logging. This proposal will specifically create a novel approach to food logging based on using a wearable sensor to detect heart rate changes associated with eating, which triggers an ecological momentary assessment (EMA) for food logging. A food log compared with already available activity logs provides enough quantitative information to understand and analyze lifestyle quantitatively for a person. This project will develop novel event mining techniques by extending machine learning approaches for building lifestyle related causal models for a person that will be used in predictive approaches as well as in guiding the person's lifestyle decisions.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.