General methods and implementations of epidemic model inference on high-performance computing platforms.

高性能计算平台上流行病模型推理的通用方法及实现。

• 批准号: 2644858
• 金额: --
• 依托单位: Lancaster University
• 依托单位国家: 英国
• 项目类别: Studentship
• 财政年份: 2022
• 资助国家: 英国
• 起止时间: 2022 至 无数据
• 项目状态: 未结题
• 来源: https://gtr.ukri.org/projects?ref=studentship-2644858
• 关键词: General; methods; implementations; epidemic; model

: Currently, parameter inference algorithms for stochastic epidemic models exist in the literature, with concrete implementations typically coded anew for each application. As demonstrated by theSARS-CoV-2 outbreak, this puts state-of-the-art inference out of reach of emergency situations when effective model fitting is required the most. This project seeks to bring together these application-specific methodologies into a general toolkit for epidemic model inference. As such, it will develop theory around stochastic state-transition models, provide improved MCMC-based fitting methods, and find novel abstractions of the underlying process models which can be reflected in a sustainable high-performance software library.

目前，随机流行病模型的参数推断算法存在于文献中，具体的实现通常为每个应用程序重新编码。正如SARS-CoV-2爆发所证明的那样，当最需要有效的模型拟合时，这使得最先进的推理无法达到紧急情况。该项目旨在将这些特定于应用程序的方法汇集到一个通用的流行病模型推理工具包中。因此，它将围绕随机状态转换模型开发理论，提供改进的基于MCMC的拟合方法，并找到可以反映在可持续高性能软件库中的底层过程模型的新抽象。