Location: This workshop will be held at the Dalla Lana School of Public Health.  

The workshop aims to inform the development of new probabilistic methods for decision support in food security. Probabilistic graphical models (PGMs) and Bayesian inference offer a framework for presenting, combining, quantifying and communicating information from multiple sources needed in decision-making involving complex (interdependent) systems. The workshop will bring together ‘experts’ in modelling and decision-making along with domain experts relating to food security research to inform the development of large-scale decision support systems (DSSs) to aid crisis management, policy design and decision-making in food security at local, regional, national and (eventually) global levels. The seven-person organising committee includes four statisticians with expertise in Bayesian methods, decision analysis, PGMs, epidemiological methods and modelling uncertainty from large-scale regional/national databases as well as three domain experts with expertise in Canada’s household food insecurity, food systems, agro-economics, environmental/occupational exposure, public and global health.

Why food security? Food security is an important driver of social and economic well-being for individuals and a nation. It involves the entire food system from farm to fork, crop production to transportation to consumption. Decision-making with regard to food security involves drawing upon different biological, physical and environmental systems as well as accounting for social, health, economic and environmental determinants and consequences. Stressors such as climate change, population growth and urbanisation, and disasters (natural/man-made) impact the sustainability of a food system at different levels and at times of the chained events, inviting risk and uncertainty and having both short and longer-term consequences. Some issues faced in combining models (of different processes) include the large diverse complex nature of data sources and types, and computational constraints of conflicting evidence or cyclical causal relationships. It is expected that this workshop and larger research outcomes will inform a new class of Bayesian Graphical Modular Models that involve combining (causal) sub-models and accounting for complex spatial and temporal variants to support decision-makers around important health, poverty and environmental problems.

Technical experts will be drawn from areas of expertise in PGMs; decision analysis and expert judgement; Bayesian methods; spatial-temporal models; multi-scale models; high dimensional data; small area estimation; and missing, sparse and extreme value data. Food domain experts will be drawn from agriculture, health, nutrition, social, economic and environmental sciences and decision-makers that complement the application of the aforementioned technical areas.

The workshop will facilitate (i) demonstrating the use of PGMs for DSS by building prototypical DSSs for food security (or aspects of it, e.g. health sub-network), (ii) informing the development of new probabilistic methods (and class of models) integrating existing techniques, for decision support in food security, and (iii) impacting a broader body of work to develop a framework for and promote evidence-informed and knowledge-based decision-making in policy and practice for complex causal systems built on sound mathematical and probabilistic principles.

Program

To view the workshop's schedule, please click here.