PIPP Phase I: Computational Foundations for Bio-social Modeling of Unseen Pandemics

大流行在社会、行为和决策环境中展开，改变了传播的地理空间模式，这取决于现有的潜在风险和适应性行为。将社会经济因素分层到传统的预测建模框架中，不足以理解这种复杂性，也不能解释人类行为和自由意志的动态和变迁。意外的人类行为起着主要作用，还有更广泛的因素，如疫苗可获得性、人类接触模式的季节性影响、病毒环境的持久性以及围绕蒙面和企业关闭的联邦和州一级政策变化。列举应该构成预测模型本身基础的有限因素清单似乎是一个巨大的挑战。该项目将推动建模作为大流行应对的一个连续、迭代和动态的组成部分，其中增量预测要强大得多，天生允许复杂性、适应性和突然性的方法有望在操作上有用。对看不见的流行病的预防需要几个相互关联的努力，涉及免疫学、机械论建模、数据驱动的建模和理解决策的社会政治背景。该项目的技术方面包括：基于机器学习的工具，用于预测病原体突变的免疫反应；基于切换动态系统的时间序列模型，用于快速适应；自适应总体采样技术；以及用于设计行为干预的模型预测控制方法。该项目将制定综合方案和框架，以a)利用前所未有的病毒病原体结合模式的技术和废水流行病学的进展，b)利用正规化模型了解预测模型在地理空间尺度上的性能变化，c)在资源受限的情况下设计有效的干预措施，d)了解其对政策制定的影响。这一奖项由跨部门的大流行预防第一阶段预测情报(PIPP)计划支持，该计划由生物科学(BIO)、计算机信息科学和工程(CEISE)、社会、行为和经济科学(SBE)和工程学(ENG)共同资助。该奖项反映了NSF的法定使命，并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，被认为值得支持。

Pandemics unfold in a social, behavioral, and decision-making context that alters the geospatial patterns of spread, depending on an existing underlying landscape of risk and adaptive behavior. Layering socioeconomic factors into traditional predictive modeling frameworks is not sufficient to understand this complexity, nor does it account for the dynamics and vicissitudes of human behavior and free-will. Unexpected human behaviors play a major role, as well as broader factors such as vaccine availability, seasonal effects from human contact patterns, viral environmental persistence, and federal and state-level policy changes around masking and business closures. Enumerating a finite list of factors that should form the basis of a predictive model itself seems like a grand challenge. This project will advance modeling as a continuous, iterative, and dynamic component of pandemic response, where incremental predictions are far more robust, and approaches that innately allow for complexity, adaptation, and surprise can be expected to be operationally useful.Pandemic prevention for unseen pandemics requires several interconnected efforts across immunology, mechanistic modeling, data-driven modeling, and understanding sociopolitical contexts of decision making. Technical aspects of the project include machine learning based tools for predicting immune response from pathogen mutations, switching dynamical systems based models of time-series for fast adaptation, adaptive population sampling techniques, and model predictive control methods for designing behavioral interventions. The project will develop integrative protocols and frameworks that a) leverage techniques for using binding patterns of pathogens for never-before-seen viruses and advances in wastewater-based epidemiology, b) understand the variation in performance of predictive models over geospatial scales using regularizing models, c) design effective interventions under resource constraints, and d) understand their impact on policy making. This award is supported by the cross-directorate Predictive Intelligence for Pandemic Prevention Phase I (PIPP) program, which is jointly funded by the Directorates for Biological Sciences (BIO), Computer Information Science and Engineering (CISE), Social, Behavioral and Economic Sciences (SBE) and Engineering (ENG).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.