complex adaptive system of systems (casos) engineering ... › casosengineering › _assets...casos...
TRANSCRIPT
Complex Adaptive System of Systems(CASoS) Engineering Initiativehttp://www.sandia.gov/CasosEngineering/
Sandia National Laboratories is a multi-program laboratory managed and operated by Sandia Corporation, a wholly owned subsidiary of Lockheed Martin Corporation, for the U.S. Department of
Energy’s National Nuclear Security Administration under contract DE-AC04-94AL85000.
Modeling Animal Movement to Help Control PPR
Walt Beyeler, Patrick Finley, Chrysm Watson Ross, William Fogleman
Animal Disease Control Meeting ‐ Dubai, UAEFebruary 16 – 20, 2013
SAND 2013‐0939C
CASoS Engineering
Animal Disease Control Meeting16‐20 February 2013
Overview
• Problem
• How a model of animal movement might help
• Model outline
• Dealing with uncertainties
- Within the model- About the model
• Questions we hope to answer during the meeting
2
CASoS Engineering
Animal Disease Control Meeting16‐20 February 2013
Economic Impact of PPR
• PPR is highly contagious, spreading quickly through flocks, with high mortality.
• Goats and sheep are sources not only of milk and meat, but also as sources of easily mobilized income, particularly in lean times and such households have very low capacity to absorb external shocks including disease outbreaks.
• PPR not only causes animal death but also leads to reductions in milk yield and weight loss.
• 62.5% of the global domestic small ruminant population is at risk.
3
CASoS Engineering
Animal Disease Control Meeting16‐20 February 2013
Assumptions
• Spread of disease is significantly influenced by animal movement
4
CASoS Engineering
Animal Disease Control Meeting16‐20 February 2013
Assumptions
• Spread of disease is significantly influenced by animal movement
• Information about animal movement can be used to help control disease spread- Monitoring specific locations for early warning- Targeting controls to conserve resources/effort- Surveilling and/or accessing known static reservoirs
5
CASoS Engineering
Animal Disease Control Meeting16‐20 February 2013
Assumptions
• Spread of disease is significantly influenced by animal movement
• Information about animal movement can be used to help control disease spread- Monitoring specific locations for early warning- Targeting controls to conserve resources/effort- Surveilling and/or accessing known static reservoirs
• We can get enough of this information, and connect it appropriately to practical actions, to help
6
CASoS Engineering
Animal Disease Control Meeting16‐20 February 2013
Characteristics of the System
• Information about incidence of infection is hard to obtain and communicate
• Diagnostic criteria might be hard to satisfy – how useful could partial or uncertain information be?
• Resources for intervening are scarce, badly distributed, and hard to deploy and apply
7
CASoS Engineering
Animal Disease Control Meeting16‐20 February 2013
Possible Controls
• Will need to be modeled themselves (rather than taken for granted) because of the special logistical difficulties in Afghanistan
• Possible kinds of controls:• Vaccination• Quarantine?• Culling? (as a control dynamic for the model, at least)?
• Constraints on administering controls that may constrain selection of a response
• Communication of decisions• Compliance• Movement of required resources (vaccines, personnel, etc.)
8
CASoS Engineering
Animal Disease Control Meeting16‐20 February 2013
Modeled Processes
9
CASoS Engineering
Animal Disease Control Meeting16‐20 February 2013
Modeled Processes
10
Movement of Diseased Animals
Movement of disease ->Movement of animals ->Movement of people
CASoS Engineering
Animal Disease Control Meeting16‐20 February 2013
Modeled Processes
11
Movement of Diseased Animals
Movement of disease ->Movement of animals ->Movement of people
Progression of infection in an animal
CASoS Engineering
Animal Disease Control Meeting16‐20 February 2013
Modeled Processes
12
Movement of Diseased Animals
Movement of disease ->Movement of animals ->Movement of people
Progression of infection in an animal
Detecting ConditionsDetection of disease (observation, diagnosis, reporting)
CASoS Engineering
Animal Disease Control Meeting16‐20 February 2013
Modeled Processes
13
Movement of Diseased Animals
Movement of disease ->Movement of animals ->Movement of people
Progression of infection in an animal
Detecting ConditionsDetection of disease (observation, diagnosis, reporting)
Transmitting InformationIncidence of disease at various locationsMay be partial, inaccurate
CASoS Engineering
Animal Disease Control Meeting16‐20 February 2013
Modeled Processes
14
Movement of Diseased Animals
Movement of disease ->Movement of animals ->Movement of people
Progression of infection in an animal
Detecting ConditionsDetection of disease (observation, diagnosis, reporting)
Transmitting InformationIncidence of disease at various locationsMay be partial, inaccurate
Planning
Responses
CASoS Engineering
Animal Disease Control Meeting16‐20 February 2013
Modeled Processes
15
Movement of Diseased Animals
Movement of disease ->Movement of animals ->Movement of people
Progression of infection in an animal
Detecting ConditionsDetection of disease (observation, diagnosis, reporting)
Transmitting InformationIncidence of disease at various locationsMay be partial, inaccurate
Planning
Responses
Deploying Information and Resources
Deploying Information and Resources
Diverse kinds of controls might be considered
CASoS Engineering
Animal Disease Control Meeting16‐20 February 2013
Movement of Diseased Animals
Transmitting InformationP
lanningR
esponses
Deploying Information and Resources
Deploying Information and Resources
Detecting Conditions
Implementing Controls
Modeled Processes
Movement of disease ->Movement of animals ->Movement of people
Progression of infection in an animal
Detection of disease (observation, diagnosis, reporting)
Incidence of disease at various locationsMay be partial, inaccurate
Diverse kinds of controls might be considered
16
CASoS Engineering
Animal Disease Control Meeting16‐20 February 2013
From Afghanistan PEACE Project’s Afghanistan Livestock Market AssessmentReport on Afghanistan Livestock Market Dynamics October 2008 – October 2009Dr. Catherine A. Schloeder and Dr Michael J. Jacobs
Space is described by a networkrather than a map
Model Structure ‐ I
Cultural practices and economics interact with
geography to shape patterns of movement.
CASoS Engineering
Animal Disease Control Meeting16‐20 February 2013
From Afghanistan PEACE Project’s Afghanistan Livestock Market AssessmentReport on Afghanistan Livestock Market Dynamics October 2008 – October 2009Dr. Catherine A. Schloeder and Dr Michael J. Jacobs
Space is described by a networkrather than a map
Model Structure ‐ I
Cultural practices and economics interact with
geography to shape patterns of movement.
Animals are moved between very different kinds of locations (pasture, village markets, major
markets). These create very different opportunities for
disease transmission.
CASoS Engineering
Animal Disease Control Meeting16‐20 February 2013
From Afghanistan PEACE Project’s Afghanistan Livestock Market AssessmentReport on Afghanistan Livestock Market Dynamics October 2008 – October 2009Dr. Catherine A. Schloeder and Dr Michael J. Jacobs
Space is described by a networkrather than a map
Model Structure ‐ I
Cultural practices and economics interact with
geography to shape patterns of movement.
Animals are moved between very different kinds of locations (pasture, village markets, major
markets). These create very different opportunities for
disease transmission.
Certain elements of geography are relevant: seasonality, security, terrain, (pasturage
value and transportability) political/tribal affiliations.
CASoS Engineering
Animal Disease Control Meeting16‐20 February 2013
Model Structure ‐ II
20
CASoS Engineering
Animal Disease Control Meeting16‐20 February 2013
Model Structure ‐ II
21
Network elements• Nodes are locations where animals stay for
significant periods (e.g. pasture, villages) or change condition (e.g. markets, butchers)
• Links are pathways between nodes. Animals’ conditions might change as they move along links
People and animals move on the network
CASoS Engineering
Animal Disease Control Meeting16‐20 February 2013
Model Structure ‐ II
22
Network elements• Nodes are locations where animals stay for
significant periods (e.g. pasture, villages) or change condition (e.g. markets, butchers)
• Links are pathways between nodes. Animals’ conditions might change as they move along links
People and animals move on the network
Transmission events depend on disease stage, animal density, location, access to treatment
Disease moves through populations at locations
CASoS Engineering
Animal Disease Control Meeting16‐20 February 2013
LatentMean duration 1.25
days
Infectious presymptomatic
Mean duration 0.5 days
IR 0.25
Infectious symptomaticCirculate
Mean duration 1.5 daysIR 1.0 for first 0.5 day,
then reduced to 0.375 for final day
Infectious symptomaticStay home
Mean duration 1.5 daysIR 1.0 for first 0.5 day,
then reduced to 0.375 for final day
Infectious asymptomaticMean duration 2 days
IR 0.25
Dead
Immune
Transition Probabilities
pS = 0.5pH = 0.5pM = 0
pHpS
(1-pS)
(1-pH)pM
pM
(1-pM)
(1-pM)
LatentMean duration 1.25
days
Infectious presymptomatic
Mean duration 0.5 days
IR 0.25
Infectious presymptomatic
Mean duration 0.5 days
IR 0.25
Infectious symptomaticCirculate
Mean duration 1.5 daysIR 1.0 for first 0.5 day,
then reduced to 0.375 for final day
Infectious symptomaticStay home
Mean duration 1.5 daysIR 1.0 for first 0.5 day,
then reduced to 0.375 for final day
Infectious asymptomaticMean duration 2 days
IR 0.25
Dead
Immune
Transition Probabilities
pS = 0.5pH = 0.5pM = 0
pHpS
(1-pS)
(1-pH)pM
pM
(1-pM)
(1-pM)
Model Structure ‐ IINetwork elements• Nodes are locations where animals stay for
significant periods (e.g. pasture, villages) or change condition (e.g. markets, butchers)
• Links are pathways between nodes. Animals’ conditions might change as they move along links
System state• Number of animals in different disease states
at each location
People and animals move on the network
Transmission events depend on disease stage, animal density, location, access to treatment
Disease progresses within animals
Disease moves through populations at locations
• Individual (representative) animals are modeled in different stages of disease.
23
CASoS Engineering
Animal Disease Control Meeting16‐20 February 2013
Uncertainties
• Information about incidence is partial and may be inaccurate. Modeling imperfections and delays in monitoring the system allows to design control measures that will work in the real environment.
• Many features of the modeled system are uncertain (disease characteristics, nomadic patterns, cross‐border interactions). Including these uncertainties in policy evaluation, and using them to guide information collection, is a central part of the approach.
24
CASoS Engineering
Animal Disease Control Meeting16‐20 February 2013
Aspirations
Define Analysis
Evaluate Performance
Define and EvaluateAlternatives
Define Conceptual Model
Satisfactory? Done
Model Development: an iterative process that uses uncertainty
CASoS Engineering
Animal Disease Control Meeting16‐20 February 2013
Aspirations
Define Analysis
Evaluate Performance
Define and EvaluateAlternatives
Define Conceptual Model
Satisfactory? Done
Model Development: an iterative process that uses uncertainty
Decision to refine the modelCan be evaluated on the sameBasis as other actions
CASoS Engineering
Animal Disease Control Meeting16‐20 February 2013
Aspirations
Define Analysis
Evaluate Performance
Define and EvaluateAlternatives
Define Conceptual Model
Satisfactory? Done
Model Development: an iterative process that uses uncertainty
Decision to refine the modelCan be evaluated on the sameBasis as other actions
Action A
PerformanceRequirement
Action B
PerformanceRequirement
CASoS Engineering
Animal Disease Control Meeting16‐20 February 2013
Aspirations
Define Analysis
Evaluate Performance
Define and EvaluateAlternatives
Define Conceptual Model
Satisfactory? Done
Model Development: an iterative process that uses uncertainty
Decision to refine the modelCan be evaluated on the sameBasis as other actions
Action A
PerformanceRequirement
Action B
PerformanceRequirement
Model uncertainty permits distinctions
CASoS Engineering
Animal Disease Control Meeting16‐20 February 2013
Aspirations
Define Analysis
Evaluate Performance
Define and EvaluateAlternatives
Define Conceptual Model
Satisfactory? Done
Model Development: an iterative process that uses uncertainty
Decision to refine the modelCan be evaluated on the sameBasis as other actions
Action A
PerformanceRequirement
Action B
PerformanceRequirement
Model uncertainty permits distinctions
Action A
PerformanceRequirement
Action B
PerformanceRequirement
CASoS Engineering
Animal Disease Control Meeting16‐20 February 2013
Aspirations
Define Analysis
Evaluate Performance
Define and EvaluateAlternatives
Define Conceptual Model
Satisfactory? Done
Action A
PerformanceRequirement
Action B
PerformanceRequirement
Decision to refine the modelCan be evaluated on the sameBasis as other actions
Model uncertainty permits distinctions
Action A
PerformanceRequirement
Action B
PerformanceRequirement
Model uncertainty obscures important distinctions, and reducing uncertainty has value
Model Development: an iterative process that uses uncertainty
30
CASoS Engineering
Animal Disease Control Meeting16‐20 February 2013
Questions for the Group
• Basic- Does animal movement contribute to the spread of disease? Or is this only believed to be the case? Does a network representation make sense?
- What role does economics play in controlling movement?- What do we want to measure (death from PPR, sale of healthy animals, income from animal sales, ….)?
- What are the specific kinds of controls we want to look at? Can animal movement be controlled in practice?
• Formulation given network structure- What are kinds of locations and processes that we should worry about first?- Should we concern ourselves with endemic (wild) population reservoirs?- Are there bottlenecks to regional animal movement?- Is there a useful SIR‐type model for PPV
• Practical- Do published data correctly reflect conditions in the ground in Afghanistan?- Do we need to access transnational data and, if so, how reliable is it?- What kinds of controls would be supported? Are there ways build support for controls that couldn’t be implemented now?
31
CASoS Engineering
Animal Disease Control Meeting16‐20 February 2013
Questions for the Group
• Basic- Does animal movement contribute to the spread of disease? Or is this only believed to be the case? Does a network representation make sense?
- What role does economics play in controlling movement?- What do we want to measure (death from PPR, sale of healthy animals, income from animal sales, ….)?
- What are the specific kinds of controls we want to look at? Can animal movement be controlled in practice?
• Formulation given network structure- What are kinds of locations and processes that we should worry about first?- Should we concern ourselves with endemic (wild) population reservoirs?- Are there bottlenecks to regional animal movement?- Is there a useful SIR‐type model for PPV
• Practical- Do published data correctly reflect conditions in the ground in Afghanistan?- Do we need to access transnational data and, if so, how reliable is it?- What kinds of controls would be supported? Are there ways build support for controls that couldn’t be implemented now?
32
CASoS Engineering
Animal Disease Control Meeting16‐20 February 2013
Questions for the Group
• Basic- Does animal movement contribute to the spread of disease? Or is this only believed to be the case? Does a network representation make sense?
- What role does economics play in controlling movement?- What do we want to measure (death from PPR, sale of healthy animals, income from animal sales, ….)?
- What are the specific kinds of controls we want to look at? Can animal movement be controlled in practice?
• Formulation given network structure- What are kinds of locations and processes that we should worry about first?- Should we concern ourselves with endemic (wild) population reservoirs?- Are there bottlenecks to regional animal movement?- Is there a useful SIR‐type model for PPV
• Practical- Do published data correctly reflect conditions in the ground in Afghanistan?- Do we need to access transnational data and, if so, how reliable is it?- What kinds of controls would be supported? Are there ways build support for controls that couldn’t be implemented now?
33
Complex Adaptive System of Systems(CASoS) Engineering Initiativehttp://www.sandia.gov/CasosEngineering/
Sandia National Laboratories is a multi-program laboratory managed and operated by Sandia Corporation, a wholly owned subsidiary of Lockheed Martin Corporation, for the U.S. Department of
Energy’s National Nuclear Security Administration under contract DE-AC04-94AL85000.
Scraps
34
CASoS Engineering
Animal Disease Control Meeting16‐20 February 2013
Food Processing
Farms
Consumers
Other Inputs
We want to use our understanding of the food processing and distribution system to inform three questions:
1) If some specified contaminant is introduced at specific location(s) what is the effect on consumers? (how many would be affected, how quickly, and how badly)
2) What are the worst plausible scenarios for question (1)
3) If some contaminant or effect is observed in the system how effectively (quickly and accurately) can we identify the points of introduction? Can quick response reduce casualties? Where else might it have gone?
4) What changes to the system (protection, information collection, etc.) would best reduce consequences and speed trace-back?
A network description which reflects processor operations, business rules, and logistical constraints, defines the possible flows of products
We want a conceptual description that is clear and grounded, but is general enough to cover a broad range of applications. This is an investment in capability and a way of transferring insights across applications.
We want a specific implementation that covers the current application without the overhead of features that might possibly be useful some day.
We want to concentrate our efforts on describing this network as fully as possible, including explicitly characterizing our epistemic and aleatory uncertainties.
35