SCH: EXP: Smart Adaptive Adherence-Enhancing Intervention Strategies for Breast Cancer Prevention

每年约有20万名妇女被诊断患有乳腺癌，超过4万人死于乳腺癌，这是美国最常见的女性癌症。晚期发现显着降低生存率;虽然早期乳腺癌的5年生存率约为97%，但晚期癌症的5年生存率仅为20%。许多临床试验和社区环境分析表明，重复使用乳房X线摄影可以显着降低乳腺癌死亡率。然而，由于筛查而导致的乳腺癌死亡率的降低取决于坚持筛查建议和连续按时进行乳房X光检查。因此，不坚持接受重复乳房X光检查的妇女有患晚期或无法治愈的乳腺癌的风险。事实上，坚持癌症筛查已被确定为降低癌症死亡率的国家首要任务。根据这一倡议，该项目的研究目标是优化自适应依从性增强干预（AEI）策略的设计和分配，以提高乳腺X线摄影筛查的总体依从性，同时减少不必要的费用。从社会的角度来看，这项研究有可能显着提高依从性增强干预策略的效率，以更有效地预防乳腺癌。这项研究的结果可以在个人健康计划、国家综合癌症控制计划以及国家一级的指南制定方面为乳腺癌预防政策提供信息。该项目还将对研究和学习的整合以及增强多样性产生直接影响。在这个项目中，一名博士生将接受培训，将系统建模方法应用于医疗保健领域。此外，研究人员还将让几名少数民族学生参与这些研究活动，旨在吸引他们从事以医疗保健为重点的工程。这项研究将应用机器学习和自适应随机动态控制方法来学习患者对增强依从性干预措施的反应，并相应地优化干预策略的使用。 如果成功的话，这个项目将做出一些智力贡献。首先，这将是第一项优化设计和分配自适应AEI策略以持续使用乳腺X射线摄影的研究。该团队将开发灵活的自适应随机控制模型，捕捉关键疾病和干预动态，对分析模型进行深入的结构分析，并开发量身定制的解决方案算法。在对这些模型进行参数化时，该小组将使用大型国家数据集为模型提供信息。此外，该团队将通过详细的模拟模型，对分析模型得出的政策与一些实际政策进行测试，以评估可能的解决方案并估计影响。该项目的方法是通用的，可以应用于其他慢性病，历史上坚持率低筛选。

Each year about 200,000 women are diagnosed with and more than 40,000 die from breast cancer, the most common female cancer in the US. Late detection significantly reduces survival; while 5-year survival is about 97 percent for early stage breast cancers, it is only about 20 percent for advanced stage cancers. Numerous clinical trials and community setting analyses have shown that repeat mammography use can significantly reduce breast cancer mortality. The reduction in breast cancer mortality due to screening however, is contingent upon adhering to screening recommendations and having consecutive on-schedule mammograms. Therefore, women who do not adhere to receiving repeat mammograms are at risk for developing advanced stage or incurable breast cancers. Indeed, adherence to cancer screening has been identified as a national top priority to reduce cancer mortality. In line with this initiative, the research objective of this project is to optimize the design and allocation of adaptive adherence-enhancing intervention (AEI) strategies to improve overall adherence to mammography screening, while reducing unnecessary costs. From a societal perspective, this research has the potential to significantly improve the efficiency of adherence-enhancing intervention strategies for more effective breast cancer prevention. Results from this research can inform breast cancer prevention policies at the level of the individual health plan, a state's comprehensive cancer control plan, and also at the national level in terms of guideline development. This project will also have an immediate impact on integration of research and learning, and enhancing diversity. Under this project, a PhD student will be trained to apply systems modeling methodologies to healthcare area. In addition, the investigators will engage several minority students into these research activities, and aim to attract them to engineering with a focus on healthcare. This research will apply machine learning and adaptive stochastic dynamic control methodologies to learn patients' responses to adherence-enhancing interventions and optimize the use of intervention strategies accordingly. If successful, this project will make several intellectual contributions. First, this will be the first study to optimize the design and allocation of adaptive AEI strategies for sustained mammography use. The team will develop flexible adaptive stochastic control models that capture key disease and intervention dynamics, conduct in depth structural analysis of analytical models, and develop tailored solution algorithms. In parameterizing such models, the team will use large national datasets to inform the models. Further, the team will test policies derived by the analytical models against some actual policies through a detailed simulation model to evaluate possible solutions and estimate the impact. The project's approaches are general and could be applied to other chronic diseases with historically low adherence rates to screening.