RAPID: CORPUS: An Edge Intelligence-Assisted Multi-Granularity COVID-19 Risk Predication and Update System

RAPID：CORPUS：边缘智能辅助的多粒度 COVID-19 风险预测和更新系统

• 批准号: 2027251
• 负责人: Weisong Shi
• 金额: $ 15万
• 依托单位: Wayne State University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2020
• 资助国家: 美国
• 起止时间: 2020-07-01 至 2022-06-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2027251&HistoricalAwards=false
• 关键词: RAPID; CORPUS; Edge; Intelligence; Assisted

Since the first COVID-19 case was diagnosed and reported at the end of December 2019, there have been more than 1.6 million COVID-19 cases reported, causing more than 100,000 death worldwide as of April 10, 2020. The outbreak of COVID-19 has significantly affected individuals and our society as a whole, and many national and international events have been canceled over COVID-19 fears, including NBA, NCAA events, Mobile World Congress. For the sake of either individual, institutes, or governments, a risk prediction and update system are urgently needed. However, to design and implement such a system, there are several system challenges. First, how to derive the infection risk level from different granularities, i.e., individual-, event-, and institution-levels? Second, how to dynamically update the risk level based on the latest outbreak news? Third, how to preserve user sensitive data while sharing adequate data for risk level calculation? To attack these challenges in this RAPID project, researchers at Wayne State University and Henry Ford Health Systems design and implement CORPUS, an edge intelligence-assisted, multi-granularity COVID-19 Risk Prediction, and Update System, which includes a mobile app running on personal phones, as well as a large-scale distributed protocol behind the app collecting and updating the information. First, CORPUS will build a multi-granularity risk analysis model, from fine-grained personal risk to small clustered meeting risk, to coarse-grained large clustered event risk, and institutional/organization risk. Second, CORPUS employs a data propagation protocol to build and update the risk analysis model. The data that can contribute to CORPUS include spatial data (such as GPS signal), temporal data (such as calendar event), as well as the input from the user (such as meeting with a specific person). Third, CORPUS leverages privacy-preserving algorithms such as node-level feature pooling and anonymous parameter of the model instead of raw user data, to ensure the confidentiality of personal information when multi-granularity models request personal risk information. With the rapid expansion of COVID-19, there is an urgent need for the individual to know their infection risk when traveling to a place in the foreseeable future. CORPUS can meet their needs by leveraging personalized information and edge intelligence. For a group or an organization, CORPUS will provide risk-related information to help them to judge the feasibility of holding a meeting or an event during an outbreak, especially for large-scale international events (such as the Olympics and World Cup). They can also proactively take action based on the risk information provided by CORPUS to reduce the spread of COVID-19. CORPUS will help governments perceive the risk of infection in their jurisdictions, and thus guide infection prevention and control for effective governance.This 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.

自2019年12月底确诊并报告首例COVID-19病例以来，截至2020年4月10日，全球已报告超过160万例COVID-19病例，造成超过10万人死亡。COVID-19的爆发对个人和我们整个社会产生了重大影响，许多国家和国际活动因COVID-19的恐惧而取消，包括NBA，NCAA活动，移动的世界大会。无论是个人、机构还是政府，都迫切需要一个风险预测和更新系统。然而，要设计和实现这样的系统，存在若干系统挑战。首先，如何从不同的粒度，即，个人、事件和机构层面？二是如何根据最新疫情消息动态更新风险等级？第三，如何保存用户敏感数据，同时共享足够的数据用于风险级别计算？为了应对这个RAPID项目中的这些挑战，韦恩州立大学和亨利福特健康系统的研究人员设计并实施了CORPUS，这是一个边缘智能辅助、多粒度的COVID-19风险预测和更新系统，其中包括一个在个人手机上运行的移动的应用程序，以及收集和更新信息的应用程序背后的大规模分布式协议。首先，CORPUS将建立一个多粒度的风险分析模型，从细粒度的个人风险到小集群会议风险，再到粗粒度的大集群事件风险，以及机构/组织风险。其次，CORPUS采用数据传播协议来建立和更新风险分析模型。对CORPUS有贡献的数据包括空间数据（如GPS信号）、时间数据（如日历事件）以及来自用户的输入（如与特定人员会面）。第三，CORPUS利用隐私保护算法，如节点级特征池和模型的匿名参数，而不是原始用户数据，以确保个人信息的机密性时，多粒度模型请求个人风险信息。随着COVID-19的迅速蔓延，个人在可预见的未来前往一个地方时迫切需要了解其感染风险。CORPUS可以通过利用个性化信息和边缘智能来满足他们的需求。对于一个团体或一个组织，CORPUS会提供与风险相关的信息，帮助他们判断在疫情期间举行会议或活动的可行性，尤其是大型国际活动（如奥运会和世界杯）。他们亦可根据CORPUS提供的风险信息主动采取行动，以减少COVID-19的传播。CORPUS将帮助各国政府了解其管辖范围内的感染风险，从而指导感染预防和控制，以实现有效的治理。该奖项反映了NSF的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

项目成果

SafeCampus: Multimodal-Based Campus-Wide Pandemic Forecasting

SafeCampus：基于多模式的校园范围内的流行病预测

• DOI: 10.1109/mic.2021.3125571
• 发表时间: 2022
• 期刊: IEEE Internet Computing
• 影响因子: 3.2
• 作者: Lu, Sidi;Wu, Baofu;Cong, Xiaoda;Yao, Yongtao;Shi, Weisong
• 通讯作者: Shi, Weisong