Building mathematical modeling workforce capacity to support infectious disease and healthcare research

项目摘要 在这个项目中，我们将培养博士生设计和进行传染病 建模分析具体目标是： (1)为时空疾病开发透明的顺序学习算法 监测和早期发现疾病群。这种基于机器学习的监控 算法以实时方式使用最新数据递归地更新其学习目标， 同时适应季节性、潜在的时空相关性和其他复杂数据 结构它不对数据分布、时空数据施加任何参数形式 变化和时空数据相关性。 (2)建立一个竞争性风险模型框架， 抗生素耐药和抗生素敏感病原体在个人层面上， 医疗中心和社区的人口水平。该框架将 医疗中心的个人暴露数据与整个社区的汇总数据， 评估遗传性，易感性和健康差异的决定因素，以及相对 卫生保健相关和社区相关感染的贡献， 环境污染和超级传播者 (3)开发一个基于代理的模型来评估1）早期结合策略的有效性 检测、抗菌干预和患者管理， 敏感和耐抗生素病原体; 2）疫苗的最佳控制策略- 可预防的传染病。这种基于代理的模型将在MInD下开发- 医疗保健框架，以提高其可重复性和普遍性。我们将系统地 评价通过监测、抗菌治疗确定的控制策略的有效性 以及在若干传输设置和突变参数下的患者管理。 该项目将产生新的统计方法，用于监视，推理和基于代理的 建模研究结果可能会影响监测实践和干预政策， 新出现的和地方性的病原体及其耐药突变体。该项目将全面 为学员在方法和实践方面进行独立和合作研究做好准备 在更广泛的背景下，与卫生保健相关的感染和疾病传播有关。

Project Summary In this project, we will train predoctoral students to design and conduct infectious disease modeling analyses. The specific aims are: (1) To develop a transparent sequential learning algorithm for spatio-temporal disease surveillance and early detection of disease clusters. This machine-learning-based surveillance algorithm recursively updates its learned objectives using up-to-date data in a real-time fashion, while accommodating seasonality, latent spatio-temporal correlation, and other complex data structure. It does not impose any parametric forms on the data distribution, spatio-temporal data variation, and spatio-temporal data correlation. (2) To develop a competing risks modeling framework for transmission dynamics of antimicrobial-resistant and antimicrobial-susceptible pathogens at the individual level in healthcare centers and at the population level in communities. This framework couples individual exposure data in healthcare centers with aggregated data in communities at large to assess transmissibility, susceptibility and health disparity determinants, and the relative contributions of healthcare-associated and community-associated infections, while accounting for environmental contamination and superspreaders. (3) To develop an agent-based model to assess 1) effectiveness of strategies combining early detection, antimicrobial intervention and patient management on containing both antimicrobial- sensitive and antimicrobial-resistant pathogens; and 2) optimal control strategies for vaccine- preventable infectious diseases. This agent-based model will be developed under the MInD- Healthcare Framework to increase its reproducibility and generalizability. We will systematically evaluate effectiveness of control strategies determined by surveillance, antimicrobial treatment and patient management under several transmission settings and mutation parameters. This project will produce novel statistical methods for surveillance, inference and agent-based modeling. Findings may potentially influence surveillance practice and intervention policies for emerging and endemic pathogens and their drug-resistant mutants. This project will fully prepare trainees for independent and collaborative research in both methodology and practice related to healthcare-associated infections and disease transmission in broader settings.