Risk Analysis of Complex and Uncertain Systems (eBook)

Preface 7
Why This Book? 7
For Whom Is It Meant? 7
What’s in It? 8
Some Specific Risk Models and Applications for Interested Specialists 12
Why Do These Models and Methods Matter? 14
Acknowledgments 15
Contents 19
Introduction to Risk Analysis 29
Quantitative Risk Assessment Goals and Challenges 30
The Quantitative Risk Assessment (QRA) Paradigm 30
Against QRA: Toward Concern-Driven Risk Management 34
Toward Less Analytic, More Pluralistic Risk Management 38
Alternatives to QRA in Recent Policy Making: Some Practical Examples 40
Concern-Driven Risk Management 42
How Effective Is Judgment-Based Risk Management? 45
Performance of Individual Judgment vs. Simple Quantitative Models 46
Performance of Consensus Judgments vs. Simple Quantitative Models 53
How Effective Can QRA Be? 58
Summary and Conclusions 59
Introduction to Engineering Risk Analysis 61
Overview of Risk Analysis for Engineered Systems 61
Using Risk Analysis to Improve Decisions 65
Hazard Identification: What ShouldWe Worry About? 65
Structuring Risk Quantification and Displaying Results: Models for Accident Probabilities and Consequences 67
Quantifying Model Components and Inputs 70
Risk Characterization 84
Methods for Risk Management Decision Making 93
Game-Theory Models for Risk Management Decision Making 96
Conclusions 98
Introduction to Health Risk Analysis 99
Introduction 99
Quantitative Definition of Health Risk 101
A Bayesian Network Framework for Health Risk Assessment 103
Hazard Identification 106
Exposure Assessment 111
Dose-Response Modeling 115
Risk and Uncertainty Characterization for Risk Management 119
Conclusions 122
Avoiding Bad Risk Analysis 124
Limitations of Risk Assessment Using Risk Matrices 125
Introductory Concepts and Examples 126
A Normative Decision-Analytic Framework 128
Logical Compatibility of Risk Matrices with Quantitative Risks 132
Risk Matrices with Too Many Colors Give Spurious Resolution 138
Risk Ratings Do Not Necessarily Support Good Resource Allocation Decisions 141
Discussion and Conclusions 146
Appendix A: Proof of Theorem 1 147
Limitations of Quantitative Risk Assessment Using Aggregate Exposure and Risk Models 149
What Is Frequency? 150
Limitations of Aggregate Exposure Metrics 157
Limitations of Aggregate Exposure-Response Models: An Antimicrobial Risk Assessment Case Study 165
Some Limitations of Risk Priority-Scoring Methods 173
Conclusions 184
Principles for Doing Better 186
Identifying Nonlinear Causal Relations in Large Data Sets 187
Nonlinear Exposure-Response Relations 188
Entropy, Mutual Information, and Conditional Independence 190
Classification Trees and Causal Graphs via Information Theory 192
Illustration for the Campylobacteriosis Case Control Data 195
Conclusions 199
Overcoming Preconceptions and Confirmation Biases Using Data Mining 201
Confirmation Bias in Causal Inferences 202
Appendix A: Computing Adjusted Ratios of Medians and Their Confidence Limits 223
Estimating the Fraction of Disease Caused by One Component of a Complex Mixture: Bounds for Lung Cancer 225
Motivation: Estimating Fractions of Illnesses Preventable by Removing Specific Exposures 225
Why Not Use Population Attributable Fractions? 226
Theory: Paths, Event Probabilities, Bounds on Causation 228
The Smoking-PAH-BPDE-p53-Lung Cancer Causal Pathway 232
Applying the Theory: Quantifying the Contribution of the Smoking- PAH- BPDE- p53 Pathway to Lung Cancer Risk 234
Uncertainties and Sensitivities 241
Discussion 242
Conclusions 243
Bounding Resistance Risks for Penicillin 245
Background, Hazard Identification and Scope: ReducingAmpicillin-Resistant E. faecium (AREF) Infections in ICUPatients 245
Methods and Data: Upper Bounds for Preventable Mortalities 247
Results Summary, Sensitivity, and Uncertainty Analysis 254
Summary and Conclusions 256
Confronting Uncertain Causal Mechanisms - Portfolios of Possibilities 259
Background: Cadmium and Smoking Risk 260
Previous Cadmium-Lung Cancer Risk Studies 261
Biological Mechanisms of Cadmium Lung Carcinogenesis 264
Quantifying Potential Cadmium Effects on Lung Cancer Risk 273
Discussion and Conclusions 279
Appendix A: Relative Risk Framework 280
Determining What Can Be Predicted: Identifiability 282
Identifiability 283
Multistage Clonal Expansion (MSCE) Models of Carcinogenesis 287
Nonunique Identifiability of Multistage Models from Input- Output Data 291
Discussion and Conclusions 296
Appendix A: Proof of Theorem 1 298
Appendix B: Listing of ITHINKTM Model Equations for the Example in Figure 11.3 300
Applications and Extensions 302
Predicting the Effects of Changes: Could Removing Arsenic from Tobacco Smoke Significantly Reduce Smoker Risks of Lung Cancer? 303
Biologically Based Risk Assessment Modeling 303
Arsenic as a Potential Human Lung Carcinogen 304
Data, Methods, and Models 307
Results 316
Sensitivities, Uncertainties, Implications, and Conclusions 318
Appendix A: Listing for TSCE Model of Smoking and Lung Cancer 320
Appendix B: Listing for MSCE Lung Cancer Model with Field Carcinogenesis 321
Simplifying Complex Dynamic Networks: A Model of Protease Imbalance and COPD Dynamic Dose- Response 323
Background on COPD 324
A Flow Process Network Model of Protease-Antiprotease Imbalance in COPD 325
Mathematical Analysis of the Protease-Antiprotease Network 328
Some Possible Implications for Experimental and Clinical COPD 333
Is the Model Consistent with Available Human Data? 334
Summary and Conclusions 336
Appendix A: Equilibrium in Networks of Homeostatic Processes 337
Value of Information (VOI) in Risk Management Policies for Tracking and Testing Imported Cattle for BSE 344
Testing Canadian Cattle for Bovine Spongiform Encephalitis ( BSE) 346
Methods and Data 349
Results 362
Discussion 365
Epilogue and Conclusions 366
Appendix: Market Impact Assumptions and Calculations 368
Improving Antiterrorism Risk Analysis 370
The Risk = Threat × Vulnerability × Consequence Framework 370
RAMCAPTM Qualitative Risk Assessment 372
Limitations of RAMCAPTM for Quantitative Risk Assessment 373
Risk Rankings Are Not Adequate for Resource Allocation 376
Some Fundamental Limitations of Risk = Threat × Vulnerability × Consequence 377
Discussion and Conclusions 386
Designing Resilient Telecommunications Networks 389
Introduction: Designing Telecommunications Infrastructure Networks to Survive Intelligent Attacks 390
Background: Diverse Routing, Protection Paths, and Protection Switching 390
A Simple Two-Stage Attacker-Defender Model 394
Results for Networks with Dedicated Routes ("Circuit-Switched" Networks) 395
Statistical Risk Models and Results for Scale-Free Packet Networks 399
Real-World Implementation Challenges: Incentives to Invest in Protection 402
Summary 406
Epilogue 407
References 409
Index 441