Plenary speakers

Mevin Hooten (Colorado State University)                                                                                                                     

President's invited lecture

Advances in animal telemetry data collection techniques have served as a catalyst for the creation of statistical methodology for analysing animal movement data. Such data and methodology have provided a wealth of information about animal space use and the investigation of individual-based animal-environment relationships. While the technology for data collection is improving dramatically over time, we are left with massive archives of historical animal telemetry data of varying quality. However, many contemporary statistical approaches for inferring movement behavior are designed for newer data that are very accurate and high-resolution. From a scientific perspective, the behaviors we are interested in learning about may be complicated, nonstationary, and constrained. Furthermore, telemetry data often contain substantial measurement error and can be nonelliptically distributed. I provide a brief overview of the history of statistical models for animal movement and present an accessible framework for accommodating the inherent characteristics of Lagrangian movement processes and uncertainty associated with telemetry data. 

Andrew Lawson (Medical University of South Carolina)

Environmental exposure assessment is a fundamental area for spatial statistical methodology. Assessment of ‘true’ environmental variation underlies much work in spatial statistics and environmetrics in general. However environmental risk estimation is an important ingredient in the estimation of spatial variation of health risk. For health outcomes that are sensitive to environmental conditions, such as respiratory/inhalation outcomes (e.g. asthma, lung cancer, bronchitis, COPD) daily exposure to environmental stressors such as atmospheric pollutants (PM2.5, NO2, SO2, black carbon, etc.) could fundamentally affect the observed incidence of these diseases. In this talk I will discuss a range of situations where environmental exposure problems lead to important methodological challenges in spatial health data analysis. 

Nathaniel Newlands (Agriculture and Agri-Food Canada) 

Humans interact with real and virtual ecosystems. Virtual (model and collaborative) ecosystems continue to expand in their knowledge, sophistication and influence in addressing increasingly complex situations and challenges involving real systems. Our world, however, continues to struggle with escalating inequality and insecurity, economic volatility, environmental resource scarcity and pollution, population growth, rapid urbanization, extreme weather, invasive species and political upheaval. Despite increasing global awareness of the urgency to address climate change and become more sustainable, societies continue to struggle in how best to transition to a low-carbon economy and take broader action aligned with sustainable development pathways. This is due to a complex array of trade-offs, varying uncertainties, changing inter-dependencies and unforeseen risks. Much of our knowledge is also domain-specific, relying strongly on historical observations of patterns and processes. To bridge this ‘knowledge-to-action’ divide, statistics has an increasingly critical role in unraveling the complexity of our world and how we construct reliable/flexible real- world solutions from interdisciplinary knowledge.

The talk will broadly cover the concept of “integrated risk” and how it may transform our current sustainable development dialogue, enabling more informed action/s. Our collective ability to sustain ecosystems and our societies in the future, over the long-term, will involve a stronger symbiosis of human and machine intelligence (‘super-intelligent tools’ that support complex decision-making).  Such tools are capable of transforming our current understanding and future capability to respond to anticipated/emergent extreme conditions and tipping-points (dynamical changes of a system’s state), in a coherent and informed way. Perspectives and recommendations on the broad application of statistics in addressing sustainable development challenges, drawing on my research within the food-water-energy nexus and agricultural sector (i.e., modeling of greenhouse-gas emissions, climate interpolation, operational forecasting, sensor-based monitoring networks, sustainability assessment), will be discussed. 

Brian Reich (North Carolina State University)                                                                                             

Secure Network sponsored lecture 

Previous studies suggest that exposure to high concentrations of air pollution during pregnancy may significantly increase the risk of birth defects and other adverse birth outcomes. While current regulations put limits on total PM2.5 concentrations, there are many speciated pollutants within this size class that likely have varying effects on perinatal health. However, due to correlations between these speciated pollutants it can be difficult to decipher their effects in a model for birth outcomes. To combat this difficulty we develop a multivariate spatio-temporal Bayesian model for the speciated particulate matter using dynamic spatial factors. These spatial factors can then be coherently interpolated to provide measurements at the pregnant mothers' homes for use in a model for birth outcomes. The introduced model for birth outcomes allows the effects of these factors to vary across different weeks of the pregnancy in order to identify susceptible periods. The proposed methodology is illustrated using pollutant monitoring data from the Environmental Protection Agency and birth records from the National Birth Defect Prevention Study. 

Kerrie Mengerson (Queensland University of Technology)                                                                                             

J Stuart Hunter Lecture 

Priors are one of the tenets of Bayesian modelling, yet their formulation, properties and impact are still a subject of much discussion. In this presentation we consider a range of 'real-world' problems in which a Bayesian approach has been instrumental. In particular, we explore the way in which the problem has informed the prior, and in return the prior has elaborated on the problem. The first example involves priors for latent variable structures, namely mixtures and hidden Markov models. The second involves elicitation and formulation of priors based on virtual reality and immersive environments. The latter problem is described in the context of conservation of threatened species, in particular rock wallabies in Australia and jaguars in the Peruvian Amazon. In addition to exploring how these priors are constructed and incorporated in the models, we ask the question: how much more insight do we gain about the problem of interest through this approach to statistical analysis?