Gamified Optimized Diabetes-management with AI powered Rural Telehealth (GODART)

项目摘要/摘要 旨在控制血糖的2型糖尿病(T2 DM)治疗循证指南 (降低的血红蛋白A1c)包括饮食、体力活动(PA)、血糖监测和 药物依附。然而，大多数T2 DM患者无法遵循这些指南 由于在初级保健环境中缺乏一致的健康行为咨询。这个问题是 在美国偏远的农村社区得到放大。作为回应，该项目旨在创建一个优化的 基于远程健康的干预-人工智能支持的电子化优化糖尿病管理 农村远程保健(Godart)。戈达特以社会认知理论为基础，将作为一种 自动化行为监控和远程教学平台。从本质上讲，Godart是一种自动化的 使用自然语言理解技术的对话式行为监控系统。在这 项目，我们建议通过利用多阶段试验和可行性测试Godart的各个组件 优化策略(MOST)。MOST是一种高效而严谨的资源管理和持续- 制定优化干预措施的改进框架。我们的建议关注的是 准备阶段，将使用完全析因试验。我们会试行和评估 评估两个不同的干预成分，每组两个水平，得出四个水平 实验条件。这些小组将测试(I)固定与自适应(游戏化)奖励计划的效果 以及(Ii)自动化与人工每周健康指导。我们将通过离职访谈来结束这个项目 与一部分参与者一起进行。研究结果将帮助我们了解交付这样一个 干预及其在降低HbA1c方面的初步有效性，导致足够有力的确证 有效性研究。

PROJECT SUMMARY/ABSTRACT Evidence-based guidelines for type 2 diabetes mellitus (T2DM) management aimed at glycemic control (reduced hemoglobin A1c) include a combination of diet, physical activity (PA), glucose monitoring, and medication adherences. However, the majority of individuals with T2DM are unable to follow these guidelines due to a lack of consistent health behavior counseling offered in the primary care setting. This problem is amplified in remote rural communities within the U.S. In response, this project aims to create an optimized telehealth-based intervention – Gamified Optimized Diabetes management with Artificial Intelligence–powered Rural Telehealth (GODART). GODART is grounded in the social cognitive theory and will serve as an automated behavior-monitoring and telecoaching platform. At the core, GODART is an automated conversational style behavior-monitoring system using natural language–understanding technologies. In this project, we propose to pilot and feasibility test the various components of GODART by leveraging multiphase optimization strategy (MOST). MOST is an efficient and rigorous resource-management and continuous- improvement framework for developing optimized interventions. Our proposal focuses on the MOST preparatory phase and will use a full factorial experimentation. We will pilot and assess the feasibility of and evaluate two different intervention components, with two levels in each of the groups, yielding four experimental conditions. These groups will test the effect of (i) a fixed vs. adaptive (gamified) rewards program and (ii) automated vs. human-delivered weekly health coaching. We will end the project with exit interviews conducted with a subset of participants. Study findings will help us learn the feasibility of delivering such an intervention and its preliminary effectiveness in reducing HbA1c, leading to adequately powered confirmatory effectiveness studies.