Technology Assisted Treatment for Binge Eating Behavior

摘要 意义：暴饮暴食行为的特征是大量进食和 伴随着失控的主观体验。认知行为疗法已被证明 是有效的，但CBT和其他暴饮暴食的选择是有限的，因为治疗并不完全有效 由于复发率和自然减少率较高。效果有限的一个关键原因是 治疗是在暴饮暴食后限制进食，这是部署CBT的好机会 旨在鼓励恢复正常进食时间表的策略。 假设：假设智能手表应用程序可以被设计用于治疗暴饮暴食障碍 使用即时适应性干预措施(JITAI)。配备复杂运动传感器和智能手表的智能手表 能够部署强大的机器学习算法的人可以被训练成被动地检测不仅是吃， 而是将一个人的暴饮暴食行为与他/她的正常饮食行为区分开来的品质。 然后，CBT策略可以在暴饮暴食后向用户展示，以鼓励患者 恢复健康的饮食模式。此外，该装置还可用于获得暴饮暴食的客观报告 剧集，以及确定每个用户的狂欢剧集的先例模式。 初步数据：调查组训练了一种能够检测进食的机器学习模型 自由生活情况，显示23/23在60小时内准确检测到的进食时间，从7个小时捕获 参与者。此外，研究还表明，两个暴饮暴食阶段在速度和效率上存在差异 在正常进食期间进食的持续时间。该团队还部署了适应性干预措施 在几个与饮食问题有关的项目中取得成功。患者和临床医生调查受访者同意 这一概念可能有助于避免暴饮暴食。 具体目标1：将改进进食检测模型，并使用以下患者的数据进行验证 照例暴饮暴食。在这一验证之后，它将在暴饮暴食的患者中部署，以确定 确定暴饮暴食的特征。具体目标2：开发智能手表应用程序作为吉泰系统，用于 交付CBT ，并咨询了专业临床医生和最终用户。 具体目标3：团队将进行 对HabitAware设备进行初步可行性、可用性和可接受性测试，以确定下一步和 第二阶段提案的进展情况。 长期目标：在开发出识别暴饮暴食的被动方法后，我们将研究 第二阶段的狂欢发作，这将允许我们进一步提前预测狂欢发作，并转移 远离患者的有害行为。我们还将扩展这款应用程序，与临床医生共享数据。

Abstract Significance: Binge-eating behavior is characterized by the consumption of a large amount of food and accompanied by the subjective experience of loss of control. Cognitive Behavioral Therapy has been shown to be effective, but CBT and other options for binge eating are limited because treatment is not effective in all patients, and because of high rates of relapse and attrition. A key reason for the limited effectiveness of treatments is restrictive eating following a binge-eating episode, which is a good opportunity to deploy CBT strategies designed to encourage a return to a normal eating schedule. Hypothesis: It is hypothesized that a smartwatch app can be designed for use in treating binge eating disorder using just-in-time adaptive interventions (JITAIs). Smartwatches with sophisticated motion sensors and capable of deploying powerful machine learning algorithms can be trained to passively detect not only eating, but the qualities that differentiate an individual’s binge eating behavior from his/her normal eating behavior. CBT strategies can then be surfaced to the user after a binge eating episode to encourage the patient to resume healthy eating patterns. Moreover, the device can be used to obtain an objective report of binge eating episodes, as well as identify each user’s patterns of antecedents to the binge episodes. Preliminary Data: The investigative team has trained a machine learning model capable of detecting eating in free living situations, showing 23/23 accurately detected eating sessions in 60 hours of data captured across 7 participants. Further, it shows that two binge-eating sessions had differentiated characteristics in rate and duration of eating from the normal eating sessions. The team also has deployed adaptive interventions successfully in several projects relating to problematic eating. Patient and clinician survey respondents agree the concept could be useful in averting binge-eating episodes. Specific Aim 1: The eating detection model will be improved and validated using data from patients who routinely binge eat. After this validation, it will be deployed across binge eating patients to determine the identifying characteristics of binge eating. Specific Aim 2: Develop the smartwatch app as a JITAI system for delivering CBT , with consultation from expert clinicians and end users. Specific Aim 3: The team will conduct an initial feasibility, usability and acceptability test of the HabitAware device to determine next steps and progress to a Phase II proposal. Long-Term Goal: After developing passive means of identifying binge eating, we will study the antecedents to binge episodes in a Phase II, which would allow us to predict binge episodes further in advance and divert the patient away from the harmful behavior. We will also extend the app to share data with a clinician.