HETEROGENEOUS AND LATENT EFFORT MODELING FOR NON-SYSTEMATIC WILDLIFE AND HUMAN HEALTH DATA

学生奖学金战略优先领域：研究领域1：生物信息学;研究领域2：数学生物学;研究领域3：统计和应用概率;研究领域4：生物学和医学中的新数学关键词：生物多样性，疾病监测，潜在模型，缺失的努力数据，统计建模摘要：生命科学在21世纪世纪面临的最大挑战之一是如何有效地利用大量可用的数据。研究人口健康和生态系统健康需要空间/时间扩展和局部详细/精确的数据。这给我们提出了一个难题，因为预算限制往往意味着科学上精确的数据在空间和时间范围上非常有限。打破这一僵局的一个有吸引力的途径是同时分析科学调查数据和机会平台数据（例如，公民科学家关于野生动物分布的报告，全科医生关于疾病流行率的记录）。这是具有挑战性的，因为这些不同的数据类型具有不同的限制。然而，它们的优势也是互补的。该项目将利用先进统计建模方面的当前思维和最新技术经验，同时分析科学数据和机会数据，实现两全其美的答案。具体而言，我们将：1.通过多种观测方法，为疾病发生率或物种丰度的遗漏或有偏见的报告过程开发正式的概率模型。量化这些不确定性来源如何受到各种时空和社会经济协变量的影响。将解释部分报告的变量与驱动潜在生物过程的变量区分开来。学生将获得现代统计建模的高度可移植性培训，并将能够在各种应用领域（传粉生态学和数字健康记录）中对尚未开发的数据集进行分析。

Studentship strategic priority area:RESEARCH AREA 1: Biological Informatics; RESEARCH AREA 2: Mathematical biology;RESEARCH AREA 3: Statistics and applied probability;RESEARCH AREA 4: New mathematics in biology and medicineKeywords: Biodiversity, Disease monitoring, Latent models, Missing effort data, Statistical modellingAbstract: One of the biggest challenges facing the life sciences in the 21st Century is the effective use of the vast quantity of data available. Studying the health of human populations and the health of ecosystems, requires data that are both spatially/temporally expansive and locally detailed/precise. This presents us with a conundrum because very often, budgetary constraints mean that scientifically precise data are highly limited in their spatial and temporal scope. One appealing route through this impasse is to analyse scientific survey data simultaneously with platform-of-opportunity data (e.g. citizen-scientist reports on the distribution of wildlife, general practitioner records on disease prevalence). This is challenging because these different data types have different limitations. However, they are also complementary in their strengths. This project will use current thinking and up-to-date technical experience in advanced statistical modelling to analyse scientific and opportunistic data simultaneously, achieving answers that make the best of both worlds. In particular, we will:1. Develop formal probabilistic models for the processes of missed, or biased reporting of disease incidence or species abundance through multiple observation methods.2. Quantify how these sources of uncertainty are affected by various spatiotemporal and socioeconomic covariates.3. Tease apart the variables that explain partial reporting, from those that drive the underlying biological process.The student will acquire highly transferrable training in modern statistical modelling and will be able to contribute to the analysis of as-yet untapped data sets in diverse and highly timely areas of application (pollinator ecology and digital health records).