CISE-MSI: RCBP-RF: SaTC: Privacy Preserving Models Leveraging Mobility Data for Public Health

该奖项全部或部分由《2021年美国救援计划法案》（公法117-2）资助。健康应用在移动的设备上的使用正变得越来越流行。随着这种普及，人们希望使用为各种公共卫生目的而生成的移动数据，例如COVID-19期间的接触者追踪。它还用于使用机器学习来推断健康风险的更复杂的应用程序。一方面，这些模式有望对有针对性的公共卫生干预措施产生变革性影响。另一方面，这些模型的结果可能会在不直接使用健康数据的情况下损害个人健康状况的隐私。即使在移动数据被去识别时，当对人的位置的物理观察增强模型的结果时，隐私也可能受到损害。被视为明显少数群体的人如果来自某种疾病发病率过高的群体，则特别脆弱。有一些方法可以调整隐私技术，帮助减轻隐私风险，但是，它们可能会损害模型的准确性。需要一种解决方案，可以产生有效的公共卫生模式，同时保护隐私。该项目的成果将是开发可以做到这一点的感染传播模型-提供准确的基于位置的数据，而不会损害健康相关应用程序的隐私。该项目将使用建立和理解有效的方法来共同设计隐私和安全技术与感染传播建模。这些隐私保护方法将通过物理观察结合对模型的查询来解释潜在的危害。我们还将为佛罗里达西北部生成一个合成人口，旨在通过数据同化进行有效更新。这种合成数据驱动的模型有可能在保护隐私的同时产生准确的结果。对研究和教育的影响将被视为在FAMU的研究能力的发展，以及通过跨学科的研究任务将由传统上在计算代表性不足的本科生进行。该项目的结果将有助于将途径扩展到计算领域和其他跨学科职业。该奖项反映了NSF的法定使命，并被认为值得通过使用基金会的智力价值和更广泛的影响审查标准进行评估来支持。

This award is funded in whole or in part under the American Rescue Plan Act of 2021 (Public Law 117-2).The use of health applications on mobile devices is becoming increasingly popular. With that popularity comes a desire to use the mobility data that is generated for various public health purposes, such as contact tracing during COVID-19. It is also used in more complex applications that use machine learning to infer health risks. On one hand, these models promise a transformative impact on targeted public health interventions. On the other hand, results from these models could compromise the privacy of an individual’s health status without directly using health data. Even when the mobility data is de-identified, privacy can be compromised when physical observations of persons’ locations augment the models’ results. People who are considered visible minorities are particularly vulnerable if they come from groups with a disproportionate prevalence of a certain disease. There are ways to adjust privacy techniques that can help mitigate privacy risks, however, they could compromise the accuracy of the models. There is a need for solutions that can yield effective public health models while preserving privacy. Results from this project will be the development of infection spread models that can do just that – give accurate place-based data without compromising privacy for health related applications. The project will use a establish and understanding of effective approaches for co-designing privacy and security techniques with infection spread modeling. These privacy protection approaches would account for potential compromise through physical observations in combination with queries to the models. We will also produce a synthetic population for Northwest Florida designed for efficient updates through data assimilation. Such synthetic-data-driven models have the potential to yield accurate results while preserving privacy. The impact on research and education will be seen in the developing of research capacity at FAMU as well as through interdisciplinary research tasks to be conducted by undergraduate students that are traditionally underrepresented in computing. Results from this project will help expand the pathways into computing fields and other interdisciplinary careersThis award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.