2. 1 Probabilistic Scenario Analysis (PSA)

3. 2 PSA -History In 1940’s - work on the atomic bomb In the 1950's - used as "what if" scenarios for nuclear proliferation (Cooke, 1991) By 1960 - financial analysis, engineering applications, and general economic evaluations Plant and animal health (Kaplan, 1993; Miller et al., 1993; McElvaine et al., 1993)

4. 3 PSA – Risk Triplet 1. What can go wrong? 2. How likely is that to happen? 3. If it does happen, what are the consequences?

5. 4 9 Points PSA Methodology State the question Identify the hazard of interest Develop a scenario tree that outlines the pathway of expected events and all the failure which could occur, culminating the occurrence of the identified hazard

6. 5 9 Points PSA Methodology Label the scenario tree and assign units Gather and document evidence Assign values to the branches of the scenario tree Perform the calculations to summarize the likelihood of the hazard occurring Consider risk management options Prepare a written report

7. 6 Linking PSA to Risk Assessment

8. 7 What can go wrong: Hazards

9. 8 What can go wrong - an outcome of hazard exposure? What can go wrong - an outcome of hazard exposure? Given a system or process with defined goals and methodologies: failures of components, procedures, safeguards, and mitigations can occur leading to hazards To determine what can go wrong: State the question to be investigated Identify the hazard of interest Develop a scenario tree

10. 9 What can go wrong - an outcome of hazard exposure? The process of defining a possible scenarios that lead to outcomes or events of interest, is called Scenario Analysis The graphic depiction of all events, successes and failures of safeguards, procedures, and components that lead to outcomes of hazard exposure is called an Event Tree, Scenario Tree, or Risk Pathway Tree

11. 10 What can go wrong - example Semen used in artificial insemination can be a means of exporting disease You are importing semen from Europe Is the semen that you receive infected? In Europe: – Herds are selected – Boars are selected from the herds – Semen is collected from the boars – The semen is then sent to you

12. 11 Scenario Tree, Event Tree Herd Infected ? Boar Infected ? Semen Infected ? Yes No Yes No Yes No Yes No Yes No Yes No Yes No Is infected semen collected? Infected Semen Collected ?

13. 12 Simplified Scenario Tree/Event Tree Herd Infected ? Boar Infected ? Semen Infected ? Yes No Yes No Yes No Is infected semen collected? Infected Semen Collected ?

14. 13 Risk Triplet 1. What can go wrong ? 2. How likely is that to happen? 3. If it does happen, what are the consequences?

15. 14 2. How likely is that to happen? A scenario tree has been developed to depict what can go wrong or right The next step is to quantify how likely it is for the hazard depicted in the scenario/event tree to occur

16. 15 2. How likely is that to happen? Identify the specific question to be answered: What is the probability of imported sperm from one animal being infected? What is the probability of imported sperm from at least 1 animal being infected? In order to answer these questions, we need to assign probabilities to the branches of each node in the tree

17. 16 Pictorial representation of probability: Event Tree A = Herd infected B = Boar infected, given herd infected C = Semen infected, given herd and boar infected Herd Infected ? Y N Boar Infected ? Semen Infected ? Y N Y N Y N Y N Y N Y N p = P(A) p = 1-P(A) p = P(B) p = 1-P(B) p = 0 p = 1 p = P(C) p = 1-P(C) p = 0 p = 1 p = 0 p = 1 p = 0 p = P(A)*P(B)*P(C) p = P(A)*P(B)*[1-P(C)] p = P(A)*[1-P(B)] p = P(A)*[1-P(B)]*0 = 0 p = [1-P(A)]*0*0 = 0 p = [1-P(A)]*1*0 = 0 p = [1-P(A)]*1*1 = 1-P(A) p = 1 - [1 - P(A)*P(B)*P(C)] n

18. 17 Pictorial representation of probability: Risk Pathway Tree-no Mitigations N Is the Herd the boar is picked from Infected ? Y Is the Boar Infected ? Y N Is the Semen Infected ? Y N Infected Semen Exported to the USA Boars Per Year from which Sperm is collected for Export Initiating Event:Decision to collect Sperm from boars for Export No Risk A = Herd infected B = Boar infected, given herd infected C = Semen infected, given herd and boar infected Prob. Imported semen from a boar is infected: p = P(A)*P(B)*P(C) Prob. Imported Semen from at least 1 boar is infected: p = 1 - [1 - P(A)*P(B)*P(C)] n Frequency of importing infected semen: f = F*P(A)*P(B)*P(C) p = P(A) p = P(B) p = P(C) F

19. 18 Pictorial representation of probability: Risk Pathway Tree- with Mitigations N Is the Herd the boar is picked from Infected ? Y Infection Detected during Inspection of Herd ? Y N Is the Boar Infected ? Y N Y N Infection Detected during pre-semen Collection Inspection of Boar ? Is the Semen Infected ? Y N Infected Semen Exported to the USA Boars Per Year from which Sperm is collected for Export Initiating Event:Decision to collect Sperm from boars for Export No Risk As Planned - No Risk No Risk As Planned - No Risk No Risk F P1 P2 P3 P4 P5 Prob. Imported semen from a boar is infected: p = P1*P2*P3*P4*P5 Prob. Imported Semen from at least 1 boar is infected: q = 1 - [1 - p] n Frequency of importing infected semen: f = F*p

20. 19 Evidence Gathering Label & Identify each parameter of the tree – F, Number of boars per year from which sperm is collected for export – P1, Probability that the herd the boar is picked from is infected – P2, Probability that infection is detected during inspection of the infected herd from which the boar is picked – P3, Probability that a boar is infected, given that it is from an infected herd that was not detected at inspection

21. 20 Evidence Gathering – P4, Probability that Infection is detected in an infected boar prior to semen collection, given that it is from an infected herd that was not detected at inspection – P5, Probability that the semen of an infected boar is infected, given that infection was not detected in the boar prior to semen collection, and given that it is from an infected herd that was not detected at inspection

22. 21 Evidence Gathering For each Node/Parameter – Gather evidence, associate it with the appropriate node/parameter, and reference it in a bibliography – Evaluate the evidence quantitatively or descriptively. Determine the min, ml, and max values of each parameter that are consistent with the available evidence

23. 22 Uncertainty in P1, Herd Prevalence Probability density function (PDF) PDF - Expresses the probability that a continuous random variable falls within some very small interval. PMF - Expresses the probability that a discrete random variable takes on a specific value. CDF - Cumulative distribution function F(x) = Prob (P1 ≤ x ) The flatter the PDF, the more the uncertainty.

24. 23 Monte Carlo Simulation A computer based methodology that uses statistical sampling techniques in obtaining a probabilistic approximation to the solution of a mathematical equation or model SymbolMINMLMAXPDF P1 P2 P3 P4 p=P1*P2*P3*P4*P5 or q = 1 - (1-p) n Result

25. 24 Risk Triplet 1. What can go wrong that? 2. How likely is that to happen? 3. If it does happen, what are the consequences?

26. 25 Pictorial representation of probability: Risk Pathway Tree-with Mitigations N Is the Herd the boar is picked from Infected ? Y Infection Detected during Inspection of Herd ? Y N Is the Boar Infected ? Y N Y N Infection Detected during pre-semen Collection Inspection of Boar ? Is the Semen Infected ? Y N Infected Semen Exported to the USA Boars Per Year from which Sperm is collected for Export Initiating Event:Decision to collect Sperm from boars for Export No Risk As Planned - No Risk No Risk As Planned - No Risk No Risk F P1 P2 P3 P4 P5 Prob. Imported semen from a boar is infected: p = P1*P2*P3*P4*P5 Prob. Imported Semen from at least 1 boar is infected: q = 1 - [1 - p] n Frequency of importing infected semen: f = F*p

27. 26 Consequences: Risk Pathway Tree Infection Caused in Sows Infected units of semen Per Year Initiating Event:Use of infected imported sperm No Risk F1 Frequency of Infection caused in sows, or in newborns: k = F2*P6 + F3*P7 N Does Sow get infected ? Y Infection Caused in Newborns Infected sows giving birth Per Year Initiating Event:Infected Sows give birth No Risk F3 N Does newborn get infected ? Y P7 P6

28. 27 Conclusion Risk Assessment Model: – allow the quantification of risk and uncertainty – help to identify gaps in knowledge, thereby defining data needs – help to standardize approaches Risk Assessment: should be transparent, flexible, documented, and consistent The assessment should effectively communicate the insights that it reveals

29. 28 Conclusion Risk assessment models: – allow the quantification of risk and uncertainty. – help to identify gaps in knowledge, thereby defining data needs. – help to standardize approaches. At a minimum: The Assessment should be transparent, flexible, documented, and consistent. The assessment should effectively communicate the insights that it reveals.