EAGER: Quantitative Modeling of Behavioral-Change for Personalized Weight Loss Interventions

肥胖是一个日益严重的问题，与糖尿病、心血管疾病和某些癌症有关。在美国，70%的成年人超重或肥胖，相关的医疗费用估计每年为1470亿美元。过去的研究表明，运动和改善饮食相结合的计划可以导致体重减轻，从而显著减少相关慢性疾病的发生。我们面临的挑战是确保人们继续参与这些劳动密集型且往往成本高昂的项目。越来越多的人相信，日益普及的数字、移动和无线技术所带来的数据和互动的可能性，有可能极大地降低成本，并增加这些项目的参与。然而，如何利用这些设备产生的数据来建模和预测行为，从而设计出有效的、个性化的、动态更新的减肥策略，这方面的知识差距仍然很大。如果获得成功，这个探索性研究早期拨款（EAGER）项目将解决这些差距，利用数字、移动和无线技术的数据作为输入，建立减肥治疗个性化的基础数学模型和计算机算法。这些基础工具有可能在未来被用于设计创新方法，以降低成本并提高减肥干预措施的功效，从而减少肥胖。该项目将开发新的方法，对通过体育活动和饮食减肥的行为变化进行定量建模，然后将模型用于个性化决策和治疗。现有的行为改变模型侧重于回顾性理解和总体水平的分析，目的是在总体水平上改善治疗；然而，个性化决策需要新的建模方法，比如本项目中将要开发的建模方法。该项目的方法是将医疗保健行为变化的定性模型，如社会认知理论，转化为数学建模框架。这些建模框架将以一种能够对行为变化的结果模型进行定量验证、比较、分析、优化和自动决策的方式进行选择。该项目的预期贡献包括：对不同的行为变化定性模型进行定量比较，确定减肥行为变化的“最佳”模型，以及针对新医疗技术和“非理性”行为模型中发现的新型不确定性的新随机规划框架。综合考虑，这些贡献将形成一个行为变化定量建模的框架，可用于设计个性化的医疗保健治疗。

Obesity is a growing problem linked with diabetes, cardiovascular disease, and some cancers. In the United States, 70 percent of American adults are overweight or obese, and related health care costs are estimated to be $147 billion annually. Past studies have shown that programs combining exercise and improved diet can lead to weight loss, and thus result in significant reduction in associated chronic illnesses. The challenge is ensuring continued participation in these labor-intensive and often expensive programs. There is a growing belief that the data and interactive possibilities of increasingly ubiquitous digital, mobile, and wireless technologies have the potential to dramatically reduce costs and increase participation in these programs. However, significant knowledge gaps remain with respect to how the data generated by these devices can be used to model and predict behavior in a way that enables the design of effective, personalized, dynamically updated weight-loss strategies. If successful, this EArly-Grant for Exploratory Research (EAGER) project will address these gaps, resulting in foundational mathematical models and computer algorithms for weight loss treatment personalization, utilizing data from digital, mobile, and wireless technologies as inputs. Such foundational tools have the potential to be used in the future to engineer innovative approaches to lower costs and increase efficacy of weight loss interventions, in order to reduce obesity. This project will lead to development of novel approaches for quantitative modeling of behavioral-change for weight loss through physical activity and diet, where the model is then used for personalized decision-making and treatments. Existing behavioral-change models focus on retrospective understanding and aggregate-level analysis for the purpose of improving treatments at an aggregate-level; however, new modeling approaches like the ones that will be developed in this project are needed for personalized decision-making. The approach of this project will be to convert qualitative models of behavioral-change in health care, such as social cognitive theory, into a mathematical modeling framework. These modeling frameworks will be chosen in a manner that will enable the quantitative validation, comparison, analysis, optimization, and automated decision-making of the resulting models of behavioral-change. Among the anticipated contributions of this project are: quantitative comparison of different qualitative models of behavioral-change, identification of a "best" model of behavioral-change for weight loss, and new stochastic programming frameworks for novel forms of uncertainty found in new medical technologies and in models with "irrational" behavior. Taken in combination, these contributions will form a framework for quantitative modeling of behavioral-change that can be used for design of personalized health care treatments.