Microtemporal Processes Underlying Health Behavior Adoption and Maintenance

摘要 成年初期(18-24岁)的特点是体重大幅增加，导致体重增加 癌症和其他慢性病的终生风险。从事足够水平的体力活动和 睡眠和限制久坐时间是预防体重增加的重要因素。然而， 从事这些健康行为的高峰期是在童年和青春期，而且 退化为初显成年期。促进体力活动、减少久坐时间的干预措施 充足的睡眠通常集中在这些行为的采用上。然而，当这些干预措施 成功的新行为模式不会被维持，通常会退回到基线水平。 传统的健康行为理论对潜在的行为因素提供的指导有限 维修。为了解决这一差距，我们的工作表明，决策和决策的双过程模型 行为可以阐明采用与维护的潜在机制的差异。 反思过程(例如，功效、商议、自我控制)可能在更大程度上在 行为领养。相反，反应性过程(例如，上下文提示、自动性、习惯)可以发挥作用 在行为维持方面发挥更大的作用。然而，反应性过程很难使用回溯来衡量 方法，因为它们可以在微时间尺度上展开(即，在几分钟或几个小时内变化)。要解决这个问题 问题，我们建议使用实时移动技术来收集密集的纵向数据 审查采用和维护的微时间进程的差异 体力活动少，久坐时间少，睡眠时间充足。我们将在内部进行一次预期的- 受试者病例--跨越12个月的交叉观察研究。种族多元化的新兴成年人 (18-24岁，N=300)将从幸福与健康队列(R01DA033296)招募。我们将进行 对反射变量的间歇性自我报告(即生态瞬时评估)；以及连续的传感器- 基于被动监控反应变量(例如，位置、社会接近程度、语音/文本通信)和 使用智能手表和智能手机的行为(即体育活动、久坐时间、睡眠)。这些数据将 用于预测受试者内部(日内、日间)行为“发作”的可能性变化 (例如，≥10分钟的体力活动，≥120分钟的久坐时间，≥7小时的睡眠)和“失误”(即未能达到 推荐水平≥7天)。具体目标是(1)具体地使用机器学习来识别 预测行为事件和行为的时变反射和反应因素的特定于人的组合 以及(2)名义上确定是否存在一般的、组级别的时变模式 预测因素，以及这些模式是否预测成功的行为维持结果。数据和 该项目的方法将有助于U01/U24密集纵向行为倡议的集体 目标是建立更具预测性的健康行为理论，为个性化干预指明目标。

SUMMARY Emerging adulthood (ages 18-24 years) is marked by substantial weight gain, leading to increased lifetime risks of cancer and other chronic diseases. Engaging in sufficient levels of physical activity and sleep, and limiting sedentary time are important contributors to the prevention of weight gain. However, engaging in these healthy behaviors peaks during the childhood and adolescent years, and steeply deteriorates into emerging adulthood. Interventions promoting physical activity, reduced sedentary time, and sufficient sleep typically focus on the adoption of these behaviors. Yet, when these interventions are successful, new patterns of behavior are not maintained and typically regress back to baseline levels. Traditional health behavior theories provide limited guidance regarding factors underlying behavior maintenance. To address this gap, our work suggests that dual-process models of decision-making and behavior can shed light on differences in the mechanisms underlying adoption versus maintenance. Reflective processes (e.g., efficacy, deliberations, self-control) may be activated to a greater extent during behavior adoption. In contrast, reactive processes (e.g., contextual cues, automaticity, habits) may play a greater role in behavior maintenance. However, reactive processes are difficult to measure using retrospective methods because they can unfold on a micro-timescale (i.e., change across minutes or hours). To solve this problem, we propose to use real-time mobile technologies to collect intensive longitudinal data examining differences in the micro-temporal processes underlying the adoption and maintenance of physical activity, low sedentary time, and sufficient sleep duration. We will conduct a prospective within- subject case-crossover observational study across a 12-month period. Ethnically-diverse, emerging adults (ages 18-24, N=300) will be recruited from the Happiness & Health Cohort (R01DA033296). We will conduct intermittent self-report (i.e., ecological momentary assessment) of reflective variables; and continuous, sensor- based passive monitoring of reactive variables (e.g., location, social proximity, voice/text communication) and behaviors (i.e., physical activity, sedentary time, sleep) using smartwatches and smartphones. These data will be used to predict within-subject variation (within-days, between-days) in the likelihood of behavior “episodes” (e.g., ≥10 min of physical activity, ≥120 min sedentary time, ≥7 hr sleep) and “lapses” (i.e., failure to achieve recommended levels ≥7 days). The specific aims are to (1) idiographically use machine learning to identify person-specific combinations of time-varying reflective and reactive factors that predict behavior episodes and lapse; and (2) nomothetically determine whether there are general, group-level patterns of time-varying predictors, and whether those patterns predict successful behavior maintenance outcomes. The data and methods from this project will contribute to the U01/U24 Intensive Longitudinal Behavior Initiative’s collective goal to build more predictive health behavior theories that specify targets for personalized interventions.