Bayesian spatial statistics for small area estimation of HIV

这项研究旨在开发贝叶斯方法用于撒哈拉以南非洲地区艾滋病毒的次国家建模和推断。需要准确的次国家估计来为减少艾滋病毒疾病负担和结束流行病所需的有针对性的政策干预提供信息。这项研究的三个关键目标。第一，建立空间随机效应模型，以更真实的视角观察空间。第二个目标是开发新的推理方法，将集成的嵌套拉普拉斯近似扩展到潜在高斯类之外的模型。最后，本研究旨在提出一种贝叶斯方法来进行调查设计，该方法将利用小区域估计模型来获得更高的样本效率。该项目属于医疗保健技术和数学科学EPSRC的研究领域。

This research aims to develop Bayesian methods for sub-national modelling and inference of HIV in sub-Saharan Africa.Accurate sub-national estimates are needed to inform the targeted policy interventions required to reduce HIV disease burden and end the epidemic.There are three key objectives of this research.The first is to create models for spatial random effects which take a more realistic view of space.The second objective is to develop new inference methods which extend the integrated nested Laplace approximation to models outside the latent Gaussian class.Lastly, this research aims to advance a Bayesian approach to survey design which will make use of the small area estimation model to achieve greater sample efficiency.This project falls within the healthcare technologies and mathematical sciences EPSRC research areas.