Analysis of Catastrophes and Their Public Health Consequences

factors and associated responses as well as the permissible operations using these scales. 

Chapter 3: Probabilistic Causation - The chapter discusses and exemplifies probabilistic causation, in the context of the advantages and main limitations of probabilistic methods. The focus is on the predictive aspects of causation, which is the essential reason for modeling cause-and-effect.

Chapter 4: Policy and Representations of Incertitude - Policy is informed by judgments, hard numbers, models of various sorts, and vague propositions that are used routinely to describe events and their possible causes. This chapter exemplifies modeling causation through fuzzy controllers and fuzzy operations (e.g., union and intersection) that aggregate membership by using aspects of established causal models of cancer, respiratory conditions, and other diseases. The importance of these to inform policy-science, relative to what might be obtained from probabilistic methods, is discussed and exemplified.
Chapter 5: Network Analysis - The chapter deals with Boolean, Bayesian, fuzzy (controllers), and other networks as means to develop graphical and mathematical aspects of causation, particularly when the disease process is deconstructed into different and heterogeneous sub-processes. Complex causation, modeled by Bayesian-probabilistic Directed Acyclic Graphs, is exemplified in the context of science-policy choices in public health. Examples from cancer and other multifactorial disease modeling are used to clarify the concepts developed in this chapter. This chapter also includes computing with fuzzy methods such as forward (data-driven search: begin with the data to produce the conclusion), and backward (goal-based search: begin with the conclusion to produce the antecedent) chaining. The advantages and disadvantages of the methods are exemplified and discussed in the context of informing science-policy rather than being extremely rigorous.

nonrandom="" influence="" opinion="" dynamics="" can="" also="" be="" modeled,="" instance,="" by="" assuming="" equal="" probability="" exposure,="" media="" attention,="" selective="" exposures.="" defensive="" avoidance,="" which="" changes="" assumed="" uniform="" distribution.="" motivated="" reasoning="" modeled="" response="" messages,="" susceptible="" refractory="" individuals,="" thus="" will="" linked="" social="" theory="" (chapter="" 9).="" are="" bottom-up="" simulations="" flexible="" identify="" emergent="" behaviors="" mechanisms.="" they="" capture="" different="" evolutionary="" changes,="" multiple="" levels="" event-producing="" exposures,="" dynamic="" learning,="" so="" on.="" contexts,="" intent="" link="" behavioral="" factors="" biological="" epidemiological="" discussed="" 7.="" although="" cannot="" extensive,="" it="" illustrates="" key="" aspects="" context="" both="" ex-post="" assessments.
Chapter 7: Multifactorial Disease Cause-and-Effect - The Evidence - The information developed in Chapters 5 and 6 are used to model the public health consequences of catastrophic events. The focus of the integration is the effects at low levels of exposure and response that are generally different from high exposure and multifactorial outcomes. Causal, disease-specific modeling should formally identify (conditional on the current state-of-knowledge) different aspects of the sets of risk factors. This chapter also discusses the effect of biases that affect epidemiological studies and the reliability of their results and describes how this book will help diminish their effect and result in more reliable knowledge. The chapter concludes with important aspects of epidemiological modeling, namely the specification error, for low exposures. This error is committed not by accounting for disease-specific risk factors and confounding variables, but rather by relying on a much simplified univariate dose-response model. 
Chapter 8: Mechanistic and Probabilistic Modeling of Events Leading to Catastrophic Consequences - The mechanisms resulting in catastrophic incidents and the subsequent analysis of the consequences of rare, large-magnitude events are discussed in terms of their steady state, equilibria, chaotic, and other trajectories. Linkages to probability distributions such as power laws are included to exemplify the differences between non-routine failure of engineered systems and natural catastrophes. The implications of an emergent system, with self-organization as the ability to develop a novel structure wholly different from the observed one (e.g., the structure of bronchi does not allow for the prediction of the whole human being), are also discussed regarding prediction to inform policy. This chapter links the probabilistic aspects of these events to their consequences as distributions with fat tails such as the stretched Pareto distribution and others, as well as discusses the possibility of warnings by observing bubbling or other behaviors prior to the unexpected consequence in the data to which a fat tail distribution is suitable. The linkages between an event and its consequence are developed using the methods discussed in the earlier chapters, because an event - e.g., a magnitude 8.0 earthquake - can generate several possible catastrophic consequences depending on factors such as time of day, state of preparedness, and others. Hence, the ability to rebound after the catastrophe requires predictive modeling of the interactions between actors or agents as well as accounting for local logistics and other states of ex-ante preparedness and the unexpected situations that generally follow the aftermath of a catastrophe. 

Chapter 9: Social Choice Theory - This chapter contains a discussion of voting and ranking of options based on the aggregation of individual expert scientific opinions. This chapter includes alternative sets of actors-stakeholders, such as would be found at a public meeting about the choice of alternative solutions to a public health problem. Aspects of decision analysis and game theory are useful to analyze these situations and are discussed, although the focus of the chapter is on how the aggregation of opinions and their criteria affect causal models adopted by individual experts and how conflicts in reaching an aggregate choice must be resolved.

Chapter 10: Integration and Fusion of Information and Knowledge in Public Health Policy-Science - The main objective of this concluding chapter is to discuss how modeling should inform public health policy-science options. An important consideration is the assessment of model quality, which can range from statistical considerations of lay individuals' evaluations to models' ex-ante prediction of a catastrophe. This aspect is examined in terms of attributes such as credibility, scientific merit, usefulness, and success stories (if available). For instance, credibility suggests aspects of scientific validity ranging from publication in high-impact journals to an official's ability to explain convincingly her/his findings to a lay audience. This chapter reviews and completes the discussions of the issues raised in the book from the aggregation of raw data to knowledge propagation and fusion.