Engaging diverse colorectal cancer survivors in the design of an adaptive text message-based intervention to improve diet quality

让不同的结直肠癌幸存者参与设计基于自适应短信的干预措施，以改善饮食质量

• 批准号: 10673783
• 负责人: Erin Lynn Van Blarigan
• 金额: $ 18.47万
• 依托单位: UNIVERSITY OF CALIFORNIA, SAN FRANCISCO
• 依托单位国家: 美国
• 财政年份: 2022
• 资助国家: 美国
• 起止时间: 2022-08-01 至 2024-06-30
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/10673783
• 关键词: Address; Affect; Age; Algorithms; Artificial Intelligence; Asian Americans; Behavior; Behavior Therapy; Black race; Cancer Etiology; Cancer Survivor; Categories; Cereals; Cessation of life; Characteristics; Colorectal Cancer; Consumption; Cultural Backgrounds; Data; Diabetes Mellitus; Diagnosis; Diet; Dietary Factors; Dietary Fiber; Dietary Intervention; Ethnic Origin; Fiber; Focus Groups; Food; Frequencies; Future; Gender; Goals; Grain; Guidelines; Health; Health behavior; Intake; Intervention; Interview; Language; Latinx; Learning; Level of Evidence; Libraries; Low Income Population; Machine Learning; Malignant Neoplasms; Mental Depression; Methods; Modeling; Motivation; Pacific Islander; Participant; Persons; Physical activity; Population Heterogeneity; Psychological reinforcement; Public Health; Questionnaires; Race; Randomized, Controlled Trials; Recommendation; Research; Role; Structure; Subgroup; Surveys; Testing; Text Messaging; Time; United States; Update; Woman; Work; acceptability and feasibility; age group; arm; behavior change; bilingualism; cancer prevention; cancer survival; caucasian American; colorectal cancer risk; design; dietary; ethnic diversity; follow-up; good diet; improved; intervention effect; learning algorithm; malignant breast neoplasm; mortality; multidisciplinary; preference; programs; racial diversity; response; text messaging intervention

Colorectal cancer (CRC) is the 2nd leading cause of cancer death in the United States. Interventions that are accessible and effective in the large, culturally diverse populations of people affected by CRC will greatly improve public health. Dietary interventions hold particular promise for reducing CRC mortality because strong evidence supports that a high-fiber diet rich in whole grains is associated with lower mortality among people with CRC. Yet, <15% of CRC survivors consume a high-fiber diet. Therefore, our goal is to develop a scalable intervention to increase intake of whole grains and reduce intake of refined grains that is acceptable to CRC survivors of varied cultural backgrounds. Text messages are a promising intervention component for reaching a wide range of CRC survivors. However, most studies have developed and tested “one size fits all” static programs with limited generalizability. To address this gap, we propose to develop a continuously adaptive text message-based intervention using reinforcement learning (RL). RL is a type of machine learning that can be used to optimize text messages over time based on participant characteristics (e.g., race/ethnicity, language), engagement, and/or behavior. Dr. Aguilera (co-I) has developed a text message intervention that uses RL to increase physical activity in low income people with depression and diabetes; messages are sent in English or Spanish. While promising, it is not known if a RL algorithm can be used to improve diet behavior. In this proposal, we will develop a RL algorithm to optimize text message content and timing based on whole grain intake assessed through text messages. Informed by the Capability Opportunity Motivation – Behavior (COM- B) model, we propose an embedded mixed method study. In Aim 1, we will conduct four focus groups among Black, Latinx, non-Latinx white, and Asian American/Pacific Islander CRC survivors (5-8/group) and 20 semi- structured interviews. We aim to understand CRC survivors’ capabilities, opportunities, and motivations for a healthy diet and preferences for text messages. Using these data, we will refine our theoretical framework and revise/update our library of >200 diet-focused text messages from previous work in non-Latinx white cancer survivors. In Aim 2, informed by Aim 1, we will develop an adaptive text message intervention using RL to increase the proportion of grains consumed that are whole grains among CRC survivors. We will determine the intervention’s feasibility and acceptability and estimate its effect on diet (intake of whole grains, refined grains, total fiber) in a 12-week single-arm study among 60 CRC survivors, approximately equal numbers of whom identify as Black, Latinx, non-Latinx white, and Asian American/Pacific Islander. Our multidisciplinary and multicultural team has the expertise needed to complete this study. Our proposal will inform a scalable dietary intervention with broad generalizability that we will ultimately test in a randomized controlled trial. Improving CRC survivors’ diet quality has potential to reduce CRC mortality and have great public health impact.

结直肠癌（CRC）是美国癌症死亡的第二大原因。干预措施是 在受儿童权利公约影响的大量、文化多样的人群中易于使用和有效将极大地 改善公共卫生。饮食干预对于降低结直肠癌死亡率尤其有希望，因为 有证据表明，富含全谷物的高纤维饮食与较低的死亡率相关 与CRC。然而，<15% 的结直肠癌幸存者食用高纤维饮食。因此，我们的目标是开发一个可扩展的 CRC 可接受的增加全谷物摄入量并减少精制谷物摄入量的干预措施 不同文化背景的幸存者。短信是一种很有前景的干预手段 范围广泛的CRC幸存者。然而，大多数研究都开发并测试了“一刀切”的静态 通用性有限的程序。为了解决这一差距，我们建议开发一种持续自适应的文本 使用强化学习（RL）进行基于消息的干预。 RL 是一种机器学习，可以 用于根据参与者特征（例如种族/民族、语言）随着时间的推移优化短信， 参与和/或行为。 Aguilera 博士（co-I）开发了一种短信干预方法，使用强化学习来 增加患有抑郁症和糖尿病的低收入人群的体力活动；消息以英语发送或 西班牙语。虽然很有希望，但尚不清楚强化学习算法是否可以用于改善饮食行为。在这个 建议，我们将开发一种基于全谷物的 RL 算法来优化短信内容和时间 通过短信评估摄入量。由能力机会动机-行为（COM- B）模型，我们提出了嵌入式混合方法研究。在目标 1 中，我们将开展四个焦点小组 黑人、拉丁裔、非拉丁裔白人和亚裔美国人/太平洋岛民 CRC 幸存者（5-8 人/组）和 20 名半幸存者 结构化面试。我们的目标是了解 CRC 幸存者的能力、机会和动机 健康饮食和短信偏好。利用这些数据，我们将完善我们的理论框架 修改/更新我们的数据库，其中包含来自之前非拉丁裔白人癌症研究的超过 200 条以饮食为中心的短信 幸存者。在目标 2 中，根据目标 1，我们将使用 RL 开发自适应短信干预 增加CRC幸存者食用全谷物的比例。我们将确定 干预的可行性和可接受性，并评估其对饮食的影响（全谷物、精制谷物的摄入量， 总纤维）在一项为期 12 周的单臂研究中，受试者为 60 名 CRC 幸存者，其中人数大致相等 识别为黑人、拉丁裔、非拉丁裔白人和亚裔美国人/太平洋岛民。我们的多学科和 多元文化团队拥有完成这项研究所需的专业知识。我们的建议将为可扩展的饮食提供信息 具有广泛普遍性的干预措施，我们最终将在随机对照试验中进行测试。改善 结直肠癌幸存者的饮食质量有可能降低结直肠癌死亡率并对公共卫生产生巨大影响。