Continual Optimization and Personalization of Just-in-Time Adaptive Interventions for SUD Prevention, Treatment, and Recovery

项目总结：项目3 在大流行后的现代世界，数字技术正在成为一种日益重要的工具 提供物质使用障碍(SUD)和艾滋病毒预防、治疗和康复服务。长期的 拟议项目的目标是使数字健康技术能够通过 前所未有的有效性和可持续性。我们建议将行为科学和 人工智能为(1)实时自适应数字健康的持续优化开发方法论 应对社会变化和不断变化的人口治疗需求的干预措施和(2)个性化 适时地根据每个人不断变化的治疗需求进行适应性数字健康干预。这将使 第二代即时自适应数字健康干预，具有增强的和高度可持续的功能 有效性。为了实现这一长期目标，我们将：(目标1)促进可持续干预的有效性 和参与度，通过集成人工智能的方法-即强化学习-来 开发随时间不断优化移动卫生干预的算法；(目标2)满足差异化 通过推广目标1算法来构建并持续优化特定于个人的移动设备，从而满足个人需求 卫生干预；(目标3)广泛测试、评估和改进目标1和目标2中开发的算法 模拟；和(目标4)传播开发的算法，以便它们可以很容易地应用于SUD/HIV 预防、治疗和康复。我们将为SUD/HIV科学家举办研讨会并出版这两本书 SUD、HIV和方法论方面的教程和新研究。我们将与传播和 培训核心开发免费、用户友好的软件，使SUD/HIV科学家能够开发、优化、 并评估他们自己的及时适应性数字健康干预措施。

PROJECT SUMMARY: PROJECT 3 In the modern, post-pandemic world, digital technology is becoming an increasingly important vehicle for the delivery of substance use disorder (SUD) and HIV prevention, treatment, and recovery services. The long-term goal of the proposed project is to enable digital health technology to deliver intervention services with unprecedented effectiveness and sustainability. We propose to integrate ideas from behavioral science and artificial intelligence to develop methodology for (1) continual optimization of just-in-time adaptive digital health interventions in response to societal changes and evolving population treatment needs and (2) personalized just-in-time adaptive digital health interventions to each individual's evolving treatment needs. This will enable a second generation of just-in-time adaptive digital health interventions with enhanced and highly sustainable effectiveness. To achieve this long-term goal, we will: (Aim 1) Promote sustainable intervention effectiveness and engagement by integrating approaches from artificial intelligence — namely reinforcement learning — to develop algorithms that continually optimize mobile health interventions over time; (Aim 2) Meet differential individual needs by generalizing Aim 1 algorithms to construct and continually optimize person-specific mobile health interventions; (Aim 3) Test, evaluate, and refine the algorithms developed in Aims 1 and 2 in extensive simulations; and (Aim 4) Disseminate the developed algorithms so that they can be readily applied in SUD/HIV prevention, treatment, and recovery. We will conduct workshops for SUD/HIV scientists and publish both tutorials and new research in SUD, HIV, and methodology venues. We will work with the Dissemination and Training Core to develop free, user-friendly software that will enable SUD/HIV scientists to develop, optimize, and evaluate their own just-in-time adaptive digital health interventions.