SABLE: Sensor-Based Assessment of Behavioral Lifestyles and Experiences

该研究项目将开发和评估SABLE，这是一种基于智能手机的工具包，用于在日常生活中不引人注目地捕捉人类行为。这些行为，如活动、社交和睡眠，与身体健康（例如，心脏病，肥胖症），精神健康（例如，抑郁、焦虑），主观幸福感（例如，情绪、压力）和表现（例如，学术、职业）。然而，当涉及到衡量真实的世界中的行为时，研究人员在衡量人们的行为方面却出奇地糟糕。问题是，在真实的世界中收集客观行为的数据几乎是不可能的，特别是如果它必须在不影响人们试图记录的行为的情况下完成。在参与者知情和同意的情况下，SABLE将使用智能手机中常规嵌入的传感器收集的数据，以高保真的方式描绘日常生活中的行为。实质上，这些数据将阐明以前未映射的人类行为的动态模式。从方法论上讲，SABLE将推进推断复杂行为的技术。SABLE将提供给研究和应用社区，使研究人员和从业人员能够以高水平的生态有效性不引人注目地测量日常行为。SABLE收集的数据将使研究人员能够解决长期存在的有关日常生活心理模式的问题，并确定预测结果的行为。SABLE将促进新的基于智能手机的干预措施，旨在通过自我洞察和行为改变来改善生活方式。该项目将融入心理学和计算机科学课程，并将为50多名本科生和研究生研究学徒提供教育，培训和指导机会。行为构成了社会和行为科学以及健康科学中许多研究的自变量或因变量，但是用于评估行为的主要技术，自我报告，受到一系列限制，从易受记忆限制和故意装病到耗时和破坏性。智能手机的出现及其普及为社会科学家收集行为数据的方式带来了革命性的机会。研究人员将使用移动传感方法来解决有关行为结构和轮廓的基本问题，因为它们在日常生活中发挥作用，并开发自动行为分类器模型（例如，独自在家看电视）。研究人员将在现有传感软件的基础上，从各种智能手机中自动收集数据，并以易于访问的格式将数据提供给社会科学研究人员。研究人员将探索有关日常生活中行为和语言表达的结构和轮廓的基本问题的答案。他们将创建一组基于传感器的分类器，研究人员可以使用这些分类器从移动传感器数据中提取行为推断，他们将为这些材料创建一个开源的在线资源。该项目通过智能手机提供自动且不受干扰地评估人们心理特征的方法，将促进行为数据在涉及人类参与者的所有研究领域和应用环境中的整合。该奖项反映了NSF的法定使命，通过使用基金会的智力价值和更广泛的影响审查标准进行评估，被认为值得支持。

This research project will develop and evaluate SABLE, a smartphone-based toolkit for unobtrusively capturing human behaviors as they unfold in the course of everyday life. These behaviors, such as activity, sociability, and sleep are associated with such significant outcomes as physical health (e.g., heart disease, obesity), mental health (e.g., depression, anxiety), subjective well-being (e.g., mood, stress), and performance (e.g., academic, occupational). When it comes to measuring behaviors in the real world, however, researchers are surprisingly bad at measuring what people do. The problem is that collecting data on objective behaviors in the real world is almost impossible to do, especially if it must be done without affecting the behavior one is trying to record. With the knowledge and consent of participants, SABLE will use data collected by sensors routinely embedded in smartphones to yield high-fidelity portraits of how behaviors are played out in everyday life. Substantively, the data will illuminate previously unmapped dynamic patterns of human behavior. Methodologically, SABLE will advance techniques for inferring complex behaviors. SABLE will be made available to the research and applied communities, allowing researchers and practitioners to measure everyday behavior unobtrusively with high levels of ecological validity. Data gathered by SABLE will allow researchers to address long-standing questions about the psychosocial patterns of day-to-day life and to identify behaviors that predict consequential life outcomes. SABLE will facilitate new smartphone-based interventions designed to improve lifestyles through self-insight and behavior change. The project will be integrated into psychology and computer science coursework, and it will provide education, training, and mentorship opportunities for more than 50 undergraduate and graduate research apprentices. Behaviors constitute the independent or dependent variables of many studies in the social and behavioral sciences and many more in the health sciences, but the predominant technology for assessing behavior, self-reports, are subject to an array of limitations ranging from susceptibility to memory constraints and deliberate malingering to being time consuming and disruptive. The advent of smartphones and their ubiquity offer an opportunity to revolutionize the way social scientists collect behavioral data. The investigators will use mobile-sensing methods to address basic questions about the structure and contours of behavior as they play out in everyday life and develop automated behavioral classifier models (e.g., watching TV alone at home) derived from smartphone sensing data. The investigators will build on existing sensing software to permit automated data collection from a broad array of smartphones, with the data delivered to social-science researchers in an easy-to-access format. The investigators will explore answers to basic questions about the structure and contours of behavior and linguistic expression in everyday life. They will create a set of sensor-based classifiers that researchers can use to extract behavioral inferences from mobile-sensor data, and they will create an open-source online resource for these materials. By providing the means to automatically and unobtrusively assess psychological characteristics from people's smartphones, this project will catalyze the integration of behavioral data in all areas of research and applied settings that involve human participants.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.