Project 3- EPI

摘要 久坐时间和衰老研究（星星）计划的项目3将调查坐着的时间， 站立时间、从坐到站的过渡和体力活动时间与健康衰老的生物标志物有关， 身体功能和5-7年死亡风险。该项目将应用新颖，准确，机器学习， 基于行为的分类方法，使用现有的加速度计数据定义新的SB变量。 将研究这些SB行为与健康衰老生物标志物的关系（横截面）， 在一个大型（>6000）、多样化（50%少数民族）队列中，到2020年的死亡率和身体功能变化 妇女健康倡议（WHI）的绝经后妇女在2012年佩戴加速度计- 2014年作为“客观身体活动和心血管健康”（OPACH）辅助研究的一部分（PI：A。 LaCroix）。在大型队列研究中，越来越多的证据表明久坐行为与 和健康，调整身体活动。此外，这些研究表明，老年人在 风险不活跃。许多研究都是横断面的，大多数都采用了自我报告的方法。 测量坐着的时间，并专注于有限的结果范围。很少有研究关注 与身体功能和保持流动性的前瞻性关系，这是非常突出的 保持老年人独立的结果。当研究包括客观 加速度计测量久坐时间和身体活动，他们采用了基于强度的 切割点。我们的研究表明，这些分界点错误地分类了包括坐在车里， 站立和行走。在项目3中，我们建议对正在研究的特定行为进行准确分类， 项目1和2，包括坐着的时间、站着的时间、从坐到站的过渡和步行/跑步 采用新的计算技术与现有的加速度计数据。我们将使用机器 学习分类器（已根据老年妇女的数据进行训练，并根据已知的事实进行分类） 准确率为85%），以评估每个行为的总分钟数。此外，我们还将完善现有的 通过观察项目1中的转换并验证新的 针对项目2中大腿佩戴倾角仪数据的算法。然后将算法应用于 项目3中现有的加速度计数据。项目3将评估与现有的 葡萄糖调节、内皮功能（炎症和血压）和脂质的生物标志物。 到2020年，将每年评估死亡率和自我报告的身体功能。模特们将探索 跨部门和长期关系的强度、形式和独立性。生物统计学核心将 在这些分析的基础上，通过探索长时间的静坐和静坐发生的时间来建立。我们 将把这些时间模式映射到项目2的结果。项目3将提供有价值的新证据， 关于如何最好地中断久坐以改善健康老龄化前景的公共卫生指南。

Abstract Project 3 of the Sedentary Time and Aging Research (STAR) Program will investigate how sitting time, standing time, sit-to-stand transitions and physical activity time are related to biomarkers of healthy aging, physical functioning and 5-7 year mortality risk. This Project will apply novel, accurate, machine-learned, behavior-based classification methods to define new SB variables using existing accelerometer data. These SB behaviors will be studied in relation to healthy aging biomarkers (cross-sectionally), and mortality and changes in physical function through 2020 in a large (>6000), diverse (50% minority) cohort of postmenopausal women from the Women's Health Initiative (WHI) who wore accelerometers in 2012- 2014 as part of the “Objective Physical Activity and Cardiovascular Health” (OPACH) ancillary study (PI: A. LaCroix). There is growing evidence in large cohort studies for a relationship between sedentary behavior and health, adjusting for physical activity. Further, these studies have shown that older adults are most at risk for inactivity. Many studies have been cross-sectional and most have employed self-reported measures of sitting time and focused on a limited range of outcomes. Few studies have focused on prospective relationships with physical functioning and maintaining mobility, which are highly salient outcomes for preserving independence in older adults. When studies have included objective accelerometer measures of sedentary time and physical activity, they have employed intensity-based cutpoints. Our research shows that these cutpoints misclassify behaviors including sitting in vehicle, standing and walking. In Project 3, we propose to accurately classify specific behaviors being studied in Projects 1 and 2, including sitting time, standing time, sit-to-stand transitions, and walking/running by employing new computational techniques with the existing accelerometer data. We will employ machine learned classifiers (that have been trained on data from older women and perform against known truths with 85% accuracy) to assess total minutes in each behavior. In addition, we will refine our existing classifiers for sit-to-stand transitions by observing transitions in Project 1 and validating the new algorithms against thigh worn inclinometer data in Project 2. The algorithms will then be applied to the existing accelerometer data in Project 3. Project 3 will assess cross-sectional relationships with existing biomarkers of glucose regulation, endothelial functioning (inflammation and blood pressure) and lipids. Mortality and self-reported physical functioning will be assessed annually through 2020. Models will explore the strength, shape and independence of relationships cross-sectionally and over time. The Biostatistics Core will build upon these analyses by exploring prolonged bouts of sitting, and the time of day when bouts occur. We will map these temporal patterns to Project 2 outcomes. Project 3 will provide valuable new evidence to inform public health guidelines on how best to interrupt sitting to improve prospects for healthy aging.