Accelerating viral outbreak detection in US cities using mechanistic models, machine learning and diverse geospatial data

摘要 自2020年1月初以来，我们的跨学科研究团队已经进行了几项研究，以阐明 新冠肺炎的新威胁并支持全美的公共卫生应对措施，导致 同行评议的出版物，在线新冠肺炎预测工具，以及与城市、州和 国家决策者。在我们与疾病预防控制中心合作开发全国建模资源的过程中 为了预防大流行，我们最近开发了一个评估多层次干预措施的国家模型 控制和缓解美国城市疫情的战略。我们通过整合以下内容使该模型适应新冠肺炎 年龄和风险组特定传播率、疾病进展、无症状的最新估计 感染和严重程度(包括住院、重症护理、通风和死亡的风险)。该模型是 旨在灵活地结合社交距离、接触者追踪-隔离、抗病毒预防 和治疗以及疫苗接种战略。 我们的补充目标是建议构建更细粒度和数据驱动的新冠肺炎模型以阐明 传播、确定高危人群、监测目标和有效控制这种情况和未来 美国城市中的流行病。目标S1：最初专注于德克萨斯州的奥斯汀-圆石城大都市区， 我们将应用这些模型来改进实时风险评估，并优化分层的时间和程度 社交疏远措施。目的S2：我们将快速评估推出抗病毒预防和 根据临床试验数据进行治疗。目标S3：我们将为我们的奥斯汀和全国车型开发用户界面 支持科学研究和公共卫生努力，以缓解新冠肺炎并为未来的大流行做好规划 威胁。这些目标与我们的母公司赠款(R01 AI151176-01)的具体目标2是协同的，在该目标中，我们 发展高分辨率的病毒传播模型以改进早期发现和控制 异常呼吸道病毒，特别是在高危人群中。

ABSTRACT Since early January 2020, our interdisciplinary research team has conducted several studies to elucidate the emerging threat of COVID-19 and support public health responses throughout the United States, resulting in peer-reviewed publications, online COVID-19 forecasting tools, and extensive engagement with city, state and national decision makers. In our collaboration with the CDC to develop a national modeling resource for pandemic preparedness, we had recently developed a national model for evaluating multi-layered intervention strategies to contain and mitigate outbreaks in US cities. We adapted the model to COVID-19 by incorporating the latest estimates for age- and risk-group specific rates of transmission, disease progression, asymptomatic infections, and severity (including risks of hospitalization, critical care, ventilation and death). The model is designed to flexibly incorporate combinations of social distancing, contact tracing-isolation, antiviral prophylaxis and treatment, as well as vaccination strategies. Our Supplementary Aims propose to build a more granular and data-driven model of COVID-19 to elucidate the transmission, identify high-risk populations, surveillance targets and effective control of this and future epidemics within US cities. Aim S1: Focusing initially on the Austin-Round Rock metropolitan area in Texas, we will apply these models to improve real-time risk assessments and optimize the timing and extent of layered social distancing measures. Aim S2: We will rapidly evaluate strategies for rolling out antiviral prophylaxis and therapy based on clinical trial data. Aim S3: We will develop user interfaces for our Austin and national models to support both scientific research and public health efforts to mitigate COVID-19 and plan for future pandemic threats. These Aims are synergistic with Specific Aim 2 of our parent grant (R01 AI151176-01), in which we are developing high-resolution models of viral transmission to improve the early detection and control of anomalous respiratory viruses, particularly in at risk populations.