System Dynamics Modeling:

Overview & Causal Loop Diagrams

Nathaniel Osgood
CMPT 394

1-17-2013
 What is System Dynamics?

A feedback-oriented perspective

A broad, evolving methodology to help

Conceptualize
Describe
Analyze
Manage
feedback systems
 System Dynamics offers
Qualitative & quantitative components
A defined, incremental and iterative modeling
process that delivers value throughout
Time-honed techniques for working with diverse
interdisciplinary stakeholders
Evolved software permitting focus on the
what is being described not how it is run
A rigorous mathematical foundation
A rich set of analysis tools
Techniques for interfacing closely with cognate
areas (e.g. statistics, decision sciences, evidence-based
practices, applied mathematics, other modeling approaches, etc.)
 Differences in Framing the Issues
Modeling methodologies are often
distinguished more fundamentally by the
questions being asked than by the answers
being given
Such methodologies will often be most
distinguished by the way in which they
frame problems
In comparing, we must be conscious of
these differences
Engagement in the Human Theatre
Uses of SD Models [Hovmand]
 [Modeling] Power to the People:
Fostering Participatory Discourse
To empower participatory modeling, System
Dynamics tends towards simpler formalisms
Capturing qualitative understanding
Easy understanding by stakeholders:
Simulation model
Inspection
Dialogue
Manipulation
Declarative (programming free) specification
Intuitive graphical representation
Features support high stakeholder involvement in
model conceptualization, formulation & analysis
 Stages of the System Dynamics
Modeling Process
A Key Deliverable!

Reference mode Parameter sensitivity Specification & Learning

Specification of reproduction analysis investigation of environm
Model scope/boundary Causal loop diagrams
intervention scenarios ents/Micr
selection. Stock & flow diagrams Parameters Matching of Cross-validation Investigation of oworlds/f
Model time horizon Policy structure intermediate time
Quantitative causal hypothetical external
Robustness&extreme case light
Identification of diagrams series
relations tests conditions simulator
key variables
Group model building Matching of s
Reference modes for Decision rules Cross-scenario
observed data pointUnit checking Normal Weight Babies Born

explanation Problem domain tests comparisons (e.g.

from Non-GDM Mother with
History of GDM
Normal Weight Babies Born
to Overweight Mothers
without GDM

Initial conditions Normal Weight Babies Born to

Mothers without GDM
Normal Weight Babies
Born from GDM

Constrain to sensible CEA)

Pregnancy

Normal Weight Babies

Born from T2DM
Pregnancy
<Birth Rate>

bounds Formal analysis

Pregnant Normal
Normal and Weight Mothers Pregnancies to
Underweight Pregnancies of with No GDM Non-Overweight Mother
Susceptibles Normal Weight
Deaths
Weight Non-Overweight History Developing GDM
Women
- Completion of Pregnancy
to Non-Overweight State

Pregnancy

Structural sensitivity
Duration Pregnant Women
Shedding Obesity Developing Persistent
Developing Obesity Overweight/Obesity

+ Normal Weight Pregnant with GDM

+ Infections
New Individuals Developing
Contacts of T2DM
Pregnancies Developing
GDM from Mother with
Pregnant GDM History
Susceptibles with

analysis
Overweight Pregnancies to Overweight
Overweight Babies Born Overweight Babies Born to Completion of Mothers with No Mother Developing GDM
Infectives + from Pregnant Overweight
Mothers
Pregnant Normal Weight
Mothers
Overweight Pregnancy to
Overweight State
GDM History
Completion of GDM
Pregnancy
+ Infectives Overweight Babies Born Overweight Babies Born from Oveweight Babies Born
Pregnancies of
Overweight
Overweight Women
from GDM Pregnancy Non-GDM Mother with from T2DM Mothers
Deaths
+ History of GDM

People Presenting Pregnant Women with GDM

that Continue on to
Overweight Individuals Developing T2DM
for Treatment Postpartum T2DM
Pregnancies for
Pregnant with
T2DM Women with GDM
History of GDM Pre-Existing History of
Women with History of
T2DM Deaths GDM
GDM Developing T2DM
Completion of Pregnancy for Mother with T2DM New Pregnancies from Completion of Non-GDM
Mother with T2DM Pregnancy for Woman with
+ History of GDM
Pregnant with
Waiting Times T2DM
Deaths from Non-T2DM
Women with History of
GDM
-

I (S I ) h I
Health Care Staff 2

h2

Qualitative &

2
Baseline
I ( S
70% 80% 90% 95% 98% 100% I ) h I
Average Variable
Cost per Cubic Meter h2
50% 60%

Semi-quantitative insights
0.6

0.45

0.3

Quantitative insights
0.15

0
0 1457 2914 4371 5828
 Value of the Modeling Process
Often the modeling process itself rather
than the models created offers the
greatest value
Modeling as theory building: Refinement of
mental models
Reflecting on
Mental models
What is & is not known / data
Different perspectives
 Benefits of Rich Stakeholder Participation
Developing rich, grounded understanding
Building community capacity
Critiquing model behavior
Fostering stakeholder cooperation
Implementing policy recommendations
Facilitating data collection design
Aiding in replanning
Keeping model updated
Empowering community self-guidance
Dignity of Risk [Hovmand]
Group Model Building [Hovmand]
Group Model Building [Hovmand]
Model Conceptualization: Feedback Loops
Loops in a causal loop diagram indicate
feedback in the system being represented
Qualitatively speaking, this indicates that a
given change kicks off a set of changes that
cascade through other factors so as to either
amplify (reinforce) or push back against
(damp, balance) the original change

Loop classification: product of signs in

loop (best to trace through conceptually)
Balancing loop: Product of signs negative
Reinforcing loop: Product of signs positive
 Example: Physical Systems
With & Without Feedbacks
Balancing
Driving / flying
Thermic regulation: Hot blooded
(homeothermy) vs. cold blooded
(ecothermic) animals

Introducing a feedback can

fundamentally alter a systems
behaviour
 Recall: A Common Problem:
Overly Narrow Mental Models
Most decisions are based on mental models
Frequently the failure to anticipate &
account for policy resistance is due to
narrow mental models
Deleterious effects are blamed on side effects
of anticipated process
Failure to consider interactions between diverse
pieces of system (each of which may be well
understood)
System dynamics seeks a broader
understanding of the underlying system
Examples of Deleterious Feedbacks
Cutting cigarette tar levels reduces cessation
Cutting cigarette nicotine levels leads to
compensatory smoking
Targeted anti-tobacco interventions lead to equally
targeted coupon programs by tobacco industry
Charging for supplies for diabetics leads to higher
overall costs by increases costs due to reduced self-
management, faster disease progression
ARVs prolong lives of HIV carriers, but lead to
resurgent HIV epidemic due to lower risk perception
Saving money by understaffing STI clinics, leads to
long treatment wait, greater risk of transmission by
infectives & bigger epidemics
Antibiotic overuse worsens pathogen resistance
Antilock breaks lead to more risky driving
Natural feedback: Intergenerational Vicious Cycles
 Causal Loop Diagram
Food Ingested

- +

Hunger

Focuses on capturing causality and

especially feedback effects
Indicates sign of causal impact (+ vs. )

y
0
x +y indicates x
y
x -y indicates 0
x
 Causal Loop Diagram
An arrow with a positive sign (+): all
else remaining equal, an increase
(decrease) in the first variable increases
(decreases) the second variable above
(below) what it would otherwise have
been.
An arrow with a negative sign (-): all
else remaining equal, an increase
(decrease) in the first variable decreases
(increases) the second variable below
(above) what it otherwise would have
been.
 Reasoning about Link Polarity
Easy to get confused regarding link
polarity in the context of a causal chain
Tips for reasoning about link polarity
for XY
Reason about this link in isolation
Do not be concerned about links preceding X or
following Y
Ask if X were to INCREASE, would Y
increase or decrease?
Increase in Y implies +,decrease in Y implies -
If answer is not clear or depends on value of X, need to
think about representing several paths between X and Y
 Tips
Variables should be noun phrases
Variables should be at least ordinal
Links should have unambiguous
polarity
Remember factors involving people
Avoid mega-diagrams
Label loops
Distinguish perceived and actual
situation
 Ambiguous Link
Ambiguous Link: Sometimes +,
sometimes -
Overtime Work Accomplished
+ per Day

Replace this by disaggregating causal

pathways by showing multiple links
Fatigue
+
Overtime -- Efficiency
+ + Work Accomplished
Greater Incorporation of
Outside Tasks at Work + per Day

+ More Time
Working
 Feedback Loops
Loops in a causal loop diagram
indicate feedback in the system being
represented
Qualitatively speaking, this indicates that a
given change kicks off a set of changes
that cascade through other factors so as
to either amplify (reinforce) or damp
(balance) original change
Loop classification: product of signs in
loop (best to trace through
conceptually)
 Stigmatization Capacity for - Employability
Productive + -
Work +
Nutrition Poverty
Health +
+ - +
- Risk of Injury & - +
Costs
Accidents +
Substance Abuse +

Dysphoria &
+
Impulse towards Stress
+
Self-Medication

Reinforcing (positive) feedback

Balancing (negative) feedback

Advantages of Recognizing Feedback
Balancing Feedback Reinforcing feedback
(Stability) (Instability)

Desirable Securing resiliency Enabling rapid intervention

Individual self- success
regulation Viral approaches, peer
messaging, rapid social change
Structure interventions (and system) to
achieve this
Undesirable Preventing policy resistance Heading off rapidly growing
& adverse lock in effects vicious cycles
Low tar & nicotine Addictions, cycle of poverty,
cigarettes SAVA
Risk perception for
Infectious diseases
and prevent this
 Gates Comments
The biggest advantage we have is that
good developers like to work with good
developers
[Cusumano&Selby,95]
Most people dont get millions of
people giving them feedback about their
productsWe have this whole group of
[2000] people in the US alone that takes
phone calls about our products and logs
everything thats done. So we have a
better feedback loop, including the
market
[Cusumano&Selby,95]
As [NT] got more applications, NT
servers got more popular. As its
gotten more popular, weve got
more applications.
Computer Reseller News, 9-23-1996
[T] he more users [the internet]
gets, the more content it gets, the
more users it gets.
Red Herring, 10-1995
Its all about scale economies and
market share. When youre
shipping a million units of Windows
software a month, you can afford to
spend $300 million a year improving
it and still sell at a low price.
[Fortune, 6-14-1993]
 Examples: Vicious/Virtuous Cycles
Positive (reinforcing) feedback can
lead to extremely rapid changes in
situation +
Word of
Mouth Sales
Likelihood of Cross Listing
+
Existing Users and Listing on Search +
Engines
+
+
Customers
New Users
Discovering Site
Length of Time Per Ease of Understanding
Download where to Make a Change
+
-
Number of Connections to + -
Music Download Server Confusing Code
+
Confusing
Likelihood of User Starting Additions
+ Multiple Simultaneous
Downloads
 Simple Causal Loops
Remaining Work
+

Change Requests +

+
Project Duration

Job Rhy thm

- +
Changes to
Aggregate
Schedule
Productivity
-
-
Willingness of Project
Particip ants to share info
with PM Estimated Design Costs Design Scope
-
beyond Target Budget
Overbearing PM
- -
M anagement Style +
+ Target Budget
PM Suspicion
 Example Balancing Loops
Balancing loops tend to be self-
regulating
+

Programs Run
Aggregate Computer Simultaneously
Responsiveness

+
-
Virtual Memory
Swapping
+
Learning from
Mistakes
Mistakes

-
 Risk Management
Schedule
Disruptions
+

-
Unmanaged Risks

Time taken for Risk

- Assessment and
Management
Longer Term Cost of Pressure:
Cutting Corners
Vicious Cycles
Turnover Vicious Cycle

- Morale -
Developer Fatigue
+

Backlog of Work -
- Resignations

Team Productivity Extra Hiring +

-- Related Work
+
Work per Team
Member
 Reinforcing Loop Dynamics:
Exponential Growth
Word of
Example +
Mouth Sales +
Likelihood of Cross Listing
+ and Listing on Search
Site Popularity Engines
+
+
Customers

Graph for Stock

20,000

Dynamic 15,000

implications 10,000

5,000

0
0 10 20 30 40 50 60 70 80 90 100
Time (M onth)

Stock : Current
 Causal Loop Dynamics: Goal Seeking
(Balancing
Treshold for Policy Loop)
Dissatisfaction to Lead to
Action Threshold Hunger to
Motivate Eating

Food Ingested
Example:
- +

Hunger
Dynamic behavior
 Causal Loop Dynamics: Oscillation
(Balancing Loop with Delay)

Causal Structure

Dynamic Behavior:
 Growth and Plateau
+

Likelihood of Cross Listing

and Listing on Search
Potential Existing Users
Engines
- Customers
Loop structure:
+

- New Users
Discovering Site

Reinforcing Loop
+ +
Word of
+ Mouth Sales -

Balancing Loop +
Internet Users Yet to
Discover Site
Customers +

Dynamic Behavior:
Graph for Customer
1 0 0 ,0 0 0

7 5 ,0 0 0

5 0 ,0 0 0

From Tsai 2 5 ,0 0 0

0
0 10 20 30 40 50 60 70 80 90 100
Time ( Mo n th )

Cu sto mer : Cu r ren t

Regulatory Mechanisms
Perverse Incentives Under Stress
 Elaborating Causal Loops
Work Pressure

+
+
Work Remaining

Productivity
-

Work Pressure
+
+
Fatigue
+
Work Remaining -
Productivity
-
 Elaborating Causal Loops 2 -
Managerial Desire
- Information
to Blame Availability
Project
Performance
+

+
Quality of Management
Decision Making

-
Managerial Desire Developer's Trust of
- +
to Blame Manager
Information
Project + Availability
Performance
+ +
Managerial Trust of
+
Developers
+

Quality of Management
Decision Making +
Exercise 1: Link & Loop Polarity
Label the polarity of each
link in this diagram Population of
Time per
Download Downloading Users

What factor is missing? Bandwidth Available

Per Connection

Is the feedback Bandwidth

Available to Server
positive or negative?
 Exercise 2: Unanticipated Side
Effects
Seeking modest investment in
Blackberries for productivity
enhancement

Muddied by unanticipated side

effects
 Exercise 3: Link & Loop Polarity

Create one or more causal loops

relating
Project Morale
Project Turnover
Workload
Project delay
Are these loops positive or negative?
Do these loops all have the same
time to kick in (time constants)?
 One Set of Loops

- Morale -
Developer Fatigue
+

Backlog of Work -
- Resignations

Team Productivity Extra Hiring +

-- Related Work
+
Work per Team
Member
 Recall: Dealing with Symptoms vs.
Causes
Our focus is often on undesirable
symptoms (high cost, schedule
delay, poor quality) rather than on
underlying causes
Often a project is in severe trouble by
the time these obvious (and easily
quantifiable) symptoms appear
Often choices of interventions fail to
consider broader (and perhaps less
quantifiable) effects of actions
A Consulting Case
 Managerial Pressure
Narrow mental model
aims for this goal

Unanticipated side effects push back

vs. time savings & cause quality
problems
 - New Hiring
A Bigger Picture Familiarity of Workers
with Codebase

Fraction of Staff that is -

Resignations
New to Project - Morale -
-- +
- -
Training Needs

Debugging Work to
Work Accomplished be Done
per Day -
+ + Quality of Released
- Product
Overtime + Fatigue
- +
+ Managerial + +
Project Lateness -
+ Pressure - Thoroughness of
+ Testing Total Patent or
- Latent Defects

Time for Bug Fix - +

Defects
Task + Resolution Rate
-
New Defect
+ - Pressure for "Quick
- Resolution Rate
Multiplexing of and Dirty" Fixes +
- Reliability -
of Bug
Tasks Fixes -
+
- Coordination and
Degree of Thought for
Fixes
 Path Dependence/Network Effects
& Lock-In
In the presence of
positive feedback,
can get lock in
effects
Similar early
conditions result
in divergent
outcomes (instability)
Example: Product largely in balance vs.
continuously in turmoil
Unstable, Critical, and Subject to
Lock-in
Software Quality
Trust
Respect
Morale
Elaborating Causal Loop Diagrams:
Most Basic Feedbacks
With Recovery (& Waning Immunity)
With Risk Perception-Driven
Behavioral Change
 Public Health System
Screening & Treatment Responses
and Vaccination As Well
Blowback: Perverse Evolutionary
& Behavioural Feedbacks
Incorporating Some Determinants of Health
 Structure as Shaping Behaviour
System structure is defined by
Stocks
Flows
Connections between them (yielding feedbacks)
Nonlinearity: The behaviour of the whole is
more than the sum of the behaviour of the
parts
Emergent behaviour would not be anticipated
from simple behaviour of each piece in turn
Stock and flow structure (including feedbacks)
of a system determines the qualitative
behaviour modes that the system can take on
(parameters determine particulars)
Changes to the feedback structure can change
behaviour in fundamental ways
 Causal Loop Structure: Dynamic Implications
Each loop in a causal loop diagram is
associated with qualitative dynamic
behavior
Most Common Single-Loop Modes of
Dynamic Behavior
Exponential growth
Goal Seeking Adjustment
Oscillation
When composed, get novel behaviors due to
shifting loop dominance
Behaviour of system more than sum of parts
e.g. Growth & Plateau, Boom & Bust, Lock-in
 Causal Loop Dynamics: Exponential Growth
(First Order Reinforcing Loop)

# of Infectives +
Individual Target
Weight
Examples + Weight Perceived
as Normal +
+
+
Mean Weight in
# New Infections Population

+ -
Dynamic implications
# of Susceptibles
 Causal Loop Dynamics: Goal Seeking
(Balancing
Treshold for Policy Loop)
Dissatisfaction to Lead to
Action Threshold Hunger to
Motivate Eating

Food Ingested
Example:
- +

Hunger
Dynamic behavior
 Causal Loop Dynamics: Oscillation
(Balancing Loop with Delay)

Causal Structure

Dynamic Behavior:
 Growth and Plateau
+

Likelihood of Cross Listing

and Listing on Search
Potential Existing Users
Engines
- Customers
Loop structure:
+

- New Users
Discovering Site

Reinforcing Loop
+ +
Word of
+ Mouth Sales -

Balancing Loop +
Internet Users Yet to
Discover Site
Customers +

Dynamic Behavior:
Graph for Customer
1 0 0 ,0 0 0

7 5 ,0 0 0

5 0 ,0 0 0

From Tsai 2 5 ,0 0 0

0
0 10 20 30 40 50 60 70 80 90 100
Time ( Mo n th )

Cu sto mer : Cu r ren t

 State of the System: Stocks
(Levels, State Variables,
Compartments)
Stocks (Levels) represent accumulations
These capture the state of the system
These can be measured at one instant in time
Stocks start with some initial value & are
thereafter changed only by flows into & out of
them
There are no inputs that immediately change stocks
Stocks are the source of delay in a system
In a stock & flow diagram, shown as rectangles
Example Model: Stocks
The Critical Role of Stocks in Dynamics

Stocks determine current state of

system
Stocks often provide the basis for
making choices
Stocks central to most disequilibria
phenomena (buildup, decay)
Lead to inertia
Give rise to delays
State Changes: Flows (Fluxes, Rates, Derivatives)
All changes to stocks occur via flows
Always expressed per some unit time: If
these flow into/out of a stock that keeps
track of things of type X (e.g. persons), the
rates are measured in X/(Time Unit) (e.g.
persons/year, $/month, gallons/second)
Typically measure over certain period of
time (by considering accumulated quantity
over a period of time)
e.g. Incidence Rates is calculated by
accumulating people over a year, revenue is
$/Time, water flow is litres/minute
Can be estimated for any point in time
Example Model: Flows
 Key Component: Stock & Flow
Flow Stock
Stock
+ Flow

Stocks
Determine
Flows
Flows
Dictate
Change in
Stocks
 Auxiliary Variables
Auxiliary variables are convenience names we
give to concepts that can be defined in terms
of expressions involving stocks/flows at
current time
Adding or eliminating an auxiliary variable does not
change the mathematical structure of the system
Critical for model transparency
Can be reused at many places
References to auxiliary variables prevents need for
modeler to think about all of details of definition
Enhanced modifiability: Single place to define
Convenient for reporting (graphing, tables) &
analyzing model dynamics
Example Model: Auxiliary
Variables
Constants & Time Series Parameters

For similar reasons to auxiliary

variables, we give names to
Model constants
Time series
Example Model: Parameters
Example System Structure, Illustrating Feedbacks
 Handling Heterogeneity
Step 1: Test (via scenarios) if heterogeneity is
likely to have substantive impact on results
Step 2: Where necessary, disaggregate
model according to heterogeneity
Small model: Duplicate stocks
Larger model: Subscripting
Step 3: Express inter-group contact patterns
via mixing & preference matrices
A Subscripted
Contacts with Infectives for
Person of Given Sex
SmokingStatus Sexual Activity
Group and Age

Contacts with Infectives

between People of Potentially
Different SmokingStatus <Per Contact Risk of
Infection (beta-f)>
<Per Contact Risk of
Infection (beta-m)>
Contacts with Infectives
between People of Potentially
Different SmokingStatus and
Ages

Male Population <Total population of females by

SmokingStatus
Transmission
Contacts with Infectives between Probability Beta
SexualActivityGroup
AgeCategory People of Potentially Different <Sexual Partner
CervicalScreeningCategory> SmokingStatus Sexual Activity Change Rate c sub
Groups and Ages keli>

<Total Infectives (f) by Age Fractional Force of Infection

SmokingStatus Sexual Activity
Group and Screening
Category>
<Total population of males by
AgeCategory SmokingStatus
Prevalence (lambda-16fb)
Force of Infection for Person of
Females Initial Total Population Initial Total Population by
Fractional Prevalence by Sex Given Sex SmokingStatus Weighted by pc and pa by
SmokingStatus Sexual Sexual Activity Group and Age Sex SmokingStatus and Sex SmokingStatus Age and
Age Sexual Activity Group
Activity Group and Age
Force of Infection
(lambda-16m) Initial Total Population
by Sex SmokingStatus
and Age

Mixing
Relative Partner
SexualActivityGroup> Acquisition Rate for
Fractional Prevalence SmokingStatus
Females Across All Relative Partner Mean partner acquisition
<Total population of males by rate for age category (c j
AgeCategory SmokingStatus Screening Groups Acquisition Rate for Age
Category (pa i) bar)
SexualActivityGroup>
Relative Partner Acquisition
Rate for Sexual Activity
Category (pc i)
Fractional
Prevalence Males
Fraction of male
population
vaccinated

Model
Fraction of Total Initial
Is Age Category Sexual Partner Change Rate c
Partnership Changes for
<Fraction of Total Initial Partnership among Ages 18 sub keli for Single-Age
Activity Group Sexual Partner Change Rate
Rate of waning from Changes for Sex Occuring with through 59 Group Categories
vaccine protection c sub keli for Ages 18
Initial fraction of People in a Given Sexual Activity through 59
(sigma-vm) Group> Total Initial Partnership
males vaccinated Changes for Sex and Activity
Group and Age Category
Total Initial Total Initial Partnership
waning
Fraction of Total Initial Partnership Changes Changes for Sex and
vaccination (m) <Total Initial Partnership
Partnership Changes for Sex for Sex Activity Group
Changes for Sex and Age
Category> Occuring with People in a Given
Age Category
Vaccinated (V-16m)
Aging of vaccinated
males ()
Age Category Mixing
Parameter Epsilon 1
Vaccination
Cumulative number of males Sexual Partner
males with infections Fraction of Contacts by someone of Change Rate c sub
keli
new male Age Category and Sex to assume that

Matrix
<Susceptible infections take place with people of each Age
(X-16m)> Category
Annual likelihood of
muV-m death vaccination for
males Age Category Identity Capital B
YearsInAgeCategory Matrix small delta ij Adjusted Sexual
(band i) Partner Change Rate
<Force of Infection Mean Time Until Age ckeflmij Total Population by Sex
(lambda-16m)> Mean Time Until Age Progression for Males <Male Mortality SmokingStatus Age and
<Female Progression by Sex
Mortality Rate by SmokingStatus Rate> Sexual Activity Group
Mortality Rate> Sex Age AgeCategory
Fraction of Total Initial Partnership
SmokingStatus Mean Duration of Total Initial Partnership Changes
Changes for Sex Occuring with for Sex and SmokingStatus and
Sexual Activity People in a Given Sexual Activity Theta
<Male Mortality Activity Group and Age Category
Rate> Group
Mixing Matrix
<Population Elements rho Total population of females
Activity Category by SmokingStatus Age and
Growth Rate q>
Mean Time Until Age Identity Matrix small Sexual Activity Group
Progression for Females delta lm Fraction of Contacts by someone of a
<Infected (Y-16m)> Per Contact Risk of specific Sexual Activity Group and Sex to <Total population of females by
assume that take place with people of SmokingStatus <Total population of males by
Infection (beta-m) SexualActivityGroup
each Sexual Activity Group AgeCategory AgeCategory SmokingStatus
CervicalScreeningCategory> SexualActivityGroup>
Age Category Mixing
Cumulative male
<Mean Time Until Age <Force of Infection Parameter Epsilon 2
infective years
Progression for Males> Aging of suceptible (lambda-16m)> Mean duration of acute Rate of waning
<Babies Born by Sex> males () Susceptible (X-16m) HPV infection in males natural immunity
infective years males
Aging of infected (sigma-zm)
males () Infected (Y-16m) Aging of immune
males() Immune (Z-16m)
Male new Entrants into
Male Children 0 to 4 Male Children 5 to 9 Male Children 10 to 11 Sexually Active Population Fraction of Total Initial Partnership
Male Aging to Age Male Aging to Age (cap beta-m) Incidence Recovery Changes for Sex Occuring with
(m) (m) Progression from People in a Given SmokingStatus
5 10 Group
Male Births Immune to Susceptible SmokingStatus
<Fraction of Population (m) Identity Matrix small
in Sexual Activity <Mean Time Until Age <Mean Time Until Age delta ef
Total Male new Category> Progression for Males>
Progression for Males>
entrants per year
Total Initial Partnership Total Initial Partnership
Deaths of Male Deaths of Male Fraction of Contacts by someone of a Changes for Sex and
Deaths of Male Changes for Sex and
Children 0 to 4 Children 5 to 9 specific SmokingStatus and Sex to SmokingStatus SmokingStatus and Activity
Children 10 to 11 Group
Male muX-m assume that take place with people of
muY-m death
Mortality Rate Years in Years 5 to 9 death muZ-m each SmokingStatus
Years in Years 0 to 4 <Fraction of children that <Fraction of Total Initial Partnership
initiate smoking by 12 years death
Years in Years Changes for Sex Occuring with
old by Sex> SmokingStatus Mixing
10 to 11 Parameter Epsilon 3 (Level of People in a Given Sexual Activity
Non-assortivity in mixing) Group>
<Male Mortality
Rate> <Male <Male Mortality
Mortality Rates for Mortality Rate> Rate>
Mortality Rates for <Total population
Mortality Rates for Children 10 to 11 by of males>
Children 0 to 4 by Sex Sex
Children 5 to 9 by Sex Total population
<Total population
Initial Children 0 Initial Children 10 of females>
to 4 by Sex Initial Children 5
to 9 by Sex to 11 by Sex

Fertility Rate per 1000

for SmokingStatus
Age Category

Fractional Prevalence Fertility Rate per Capita

Among Females for SmokingStatus Age
Category <Fraction of Babies who
Fractional Prevalence of are of a given sex>
Infection in Females by Population
Total Infectives (f) Age Growth Rate q
Total Infectives
(f) by Age
Total Infectives (f) Fractional Prevalence Babies Born to Mother in
Babies Born
Female Population Detected CIN1 Rate
by SmokingStatus Among Females by
SmokingStatus
<Total Infectives
(f) by Age>
Total Population of
females by AgeCategory
SmokingStatus and
AgeCategory
<Total Infectives
Babies Born by Sex

(f) by Age>
(kappa-16fbis1) Total Infectives (f)
by SmokingStatus Total population
Age
Detected CIN2 Rate of females
(kappa-16fbis2) Total Infectives (f) by
Liquid-based cytology SmokingStatus Sexual
sensitivity (papsn s) for Activity Group Total Population of
CIN1 Routine Screening coverage by females by
Detected CIN3 Rate Total population of SmokingStatus
Total population Age and Cervical Screening females by AgeCategory
by (kappa-16fbis3) SmokingStatus
SmokingStatus Category (cover i) Total Infectives (f) by
SmokingStatus Sexual
Liquid-based cytology Activity Group Age
sensitivity (papsn s) for Detected CIS1 Rate <Treated and Infected
CIN2 CIN3 CIS1 CIS2 (kappa-16fbis4) Carcinoma s1
Total population
Total population of females <Treated and Infected (ICIS1-16fb)>
of males <Undetected CIN1

Totals
(CIN1-16fb)> <Undetected by SmokingStatus Carcinoma s2
Detected CIS2 Rate <Treated and Cured
Carcinoma s1 SexualActivityGroup (ICIS2-16fb)>
(kappa-16fbis5) TCINs (TCINs-16fb)>
<Undetected CIN2(CIS1-16fb)> <Treated and Cured <Detected Carcinoma s1
Total population of <Total population of Total population of females
(CIN2-16fb)> TCISs (TCISs-16fb)> (DCIS1-16fb)>
males by females by by SmokingStatus
SmokingStatus> <Undetected <Undetected Carcinoma s2
<mu-CCD <mu-DCCL SmokingStatus SexualActivityGroup
<Detected CIN1 Carcinoma s2 (CIS2-16fb)>
<mu-CCR death> death> (DCIN1-16fb)> AgeCategory
<Susceptible <Undetected CIN3 (CIS2-16fb)> <Treated and Infected <Undetected Carcinoma s1
death> <mu-DCCR Carcinoma s1 <Treated and Infected
(X-16m)> (CIN3-16fb)> (CIS1-16fb)> <Detected Carcinoma s2
death> (ICIS1-16fb)> CIN1 (ICIN1-16fb)>
Total population of males Male Total Deaths
(DCIS2-16fb)>
<mu-ICIN1 by SmokingStatus <Detected CIN2 <Treated and Infected
<mu-CCL <mu-DCIS1 SexualActivityGroup
death> (DCIN2-16fb)> CIN3 (ICIN3-16fb)>
death> death> <Treated and Infected
<mu-ICIN2 <mu-DCCD <muV-m death>
Total population of males by Carcinoma s2 <Detected CIN1 <Undetected CIN1
death> <mu-DCIS2 death> AgeCategory SmokingStatus (ICIS2-16fb)> <Treated and Infected (DCIN1-16fb)> (CIN1-16fb)>
death> <Infected (Y-16m)>
SexualActivityGroup <Detected CIN3 CIN2 (ICIN2-16fb)>
<muZ-m death> (DCIN3-16fb)>
<mu-ICIN3 <Detected CIN2
<mu-DCIN3 Total Infectives (f) by Age
death> <muX-m death> <Detected Distant Total population of females by (DCIN2-16fb)>
death> SmokingStatus Sexual Activity <Treated and Infected <Undetected CIN2
Cumulative number of Group and Screening Category (DCCD-16fb)> SmokingStatus
CIN1 (ICIN1-16fb)> SexualActivityGroup (CIN2-16fb)>
female with infections <muY-m death> <Detected Carcinoma
AgeCategory
<Immune (Z-16m)> <Vaccinated s1 (DCIS1-16fb)> <Treated and Cured CervicalScreeningCategory
(V-16m)> TCISs (TCISs-16fb)>
<mu-ICIS1 death> <Undetected CIN3
<mu-DCIN2 (CIN3-16fb)>
death> <Detected Carcinoma
new female <Detected CIN3
s2 (DCIS2-16fb)>
infections <Treated and Infected <Undetected Distant (DCIN3-16fb)>
<mu-SCC Female Total <Undetected Local
<Susceptible CIN2 (ICIN2-16fb)> <Treated and Cured TCINs (CCD-16fb)>
death> Deaths (CCL-16fb)>
(X-16fb)> <muX-fb (TCINs-16fb)>
<mu-ICIS2 <Detected Local <Undetected Local <Detected Regional
death> death> (DCCL-16fb)> (CCL-16fb)> (DCCR-16fb)>
<mu-DCIN1
death>
<Undetected
<Detected Regional Regional (CCR-16fb)> <Vaccinated
(V-16fb)> <Detected Local
(DCCR-16fb)> <Treated and Infected <Undetected Regional (DCCL-16fb)>
<Cancer Survivors
<Force of Infection <mu-TCISs CIN3 (ICIN3-16fb)> (CCR-16fb)>
(SCC)>
(lambda-16fb)> death> <mu-CIN1 <mu-CIS2 <Undetected Distant <Infected
<muY-fb <Detected Distant
death> death> <Infected (Y-16fb)> (CCD-16fb)> <Immune (Z-16fb)> (Y-16fb)>
death> (DCCD-16fb)>
<mu-CIN2 <Cancer
death> Survivors (SCC)> <Susceptible
<mu-CIN3 <muZ-fb death> (X-16fb)>
<mu-TCINs death> <mu-CIS1
death> death>
<muV-fb death>

Male
Initial Population by Sex
SmokingStatus AgeCategory
Fraction of CINs
regressions clearing CIN
that also clear infection
(gamma bar-16f)

Initial Population by Sex

SmokingStatus
Cumulative female Fraction of Population in SexualActivityCategory and
infective years Sexual Activity Category AgeCategory Rate of Recovery from
Recovered from Treated & Infected CIN1 infection wtih HPV
infective years (gamma-16f)
<Infected (Y-16fb)> females

Initial population females (N-f) by Recovered from Detected CIN1

AgeCategory SmokingStatus <Regression CIN1 to
Initial population of
SexualActivityGroup Infected Rate
males (N-m) (tau-16fbs1)>
CervicalScreeningCategory

Recovered from Undetected CIN1

Fraction of Females never
undergoing cervical screening <Infected (Y-16fb)>

Cu
category (tadpole b)
<Female
Mortality Rate>

Population
<Total population of females by Initial population females (N-f)
SmokingStatus by AgeCategory
SmokingStatus
SexualActivityGroup SexualActivityGroup
AgeCategory <Female
CervicalScreeningCategory> Fraction of Females entering
cervical screening category Mortality Rate> mu-DCIN1
Recovery (f) death
(tadpole b) Mean duration of acute HPV
infection females
Fraction of female
Initial Infected
population vaccinated mu-CIN1
Per Contact Risk
of Infection death
<Fractional (beta-f)
Recurrence of CIN1
Prevalence Males> (theta-rs1)
Fraction of Initial
<Fraction of Population Population that Starts <Female
in Sexual Activity <Female Infective Mortality Rate>
Mortality Rate> <Mean Time Until Age
Category> Progression for Females> Regression from Treated & Infected CIN1 to
muX-fb Undetected CIN1
<Fraction of Females Undetected
death CIN1 Rate <Detected CIN1 Rate
entering cervical screening <Force of Infection <Mean Time Until Age
(lambda-16fb)> (theta-16fbs1) (kappa-16fbis1)> mu-ICIN1
category (tadpole b)> Progression for Females>
Rate of waning from death
vaccine protection <Mean Time Until Age % CIN1 infected
(sigma-vf) after treament
Progression for Females> (psi-s1)
Cure Rate of CIN1
Aging of undetected Progression from Undetected
(cap-gamma-s1)
<Mean Time Until Age Aging of susceptible Aging of infected CIN1 () CIN1 to Detected CIN1
Progression for Females> females () females ()
Susceptible (X-16fb) Detected CIN1 (DCIN1-16fb) Aging of treated
Undetected CIN1 (CIN1-16fb)
and infected CIN1 Treated and Infected CIN1 (ICIN1-16fb)
Infected (Y-16fb) Progression from Infected to Aging of
Aging of vaccinated Incidence
Vaccinated (V-16fb) waning vaccination (f) Undetected CIN1 detected CIN1 ()
females ()
Progression from
Vaccination Regression from Undetected Detected CIN1 to
muY-fb Treated & Infected CIN1
females death CIN1 to Infected <Mean Time Until Age
<Mean Time Until Age
<Female Progression for Females>
<Regression CIN1 to Progression for Females>
Mortality Rate>
Progression from Infected to Undetected CIN2 Infected Rate Regression from
<Female
(tau-16fbs1)> Detected CIN1 to
Mortality Rate>
Infected
muV-fb death Annual likelihood of
vaccination for
females Regression CIN1 to Infected
<Years in Years Progression from
Rate (tau-16fbs1)
10 to 11> Progression from Detected CIN1 to
Initial fraction of Progression from
<YearsInAgeCategory Detected CIN1 to Treated & Cured TCINs
females vaccinated Fraction of children that Undetected Undetected CIN1 to Undetected CIN2
(band i)> CIN2 Rate
initiate smoking by 12 years Progression Undetected CIN2
<Babies Born by Sex> old by Sex from Infected (theta-16fbs2)
to Undetected <Regression CIN2 to Recurrence of CIN2
CIN3 Undetected Infected Rate (theta- rs2) <% CIN1 infected after
Regression from
CIN1 Rate (tau-16fbs2)> treament (psi-s1)>
Total Female new (pi-16fbs1) Detected CIN2 to
Entrants per Year Infected
Undetected <Cure Rate of CIN1
CIN3 Rate (cap-gamma-s1)>
Female Children 0 to 4 Female Children 5 to 9 Female Children 10 to 11 <Female Total Female new
Female Aging to (theta-16fbs3)
Female Aging to Deaths> Entrants flow
Female Births Age 5
Age 10 Regression from
Undetected CIN3 to
<Mean Duration of <Mean Time Until Age
Infected Regression from
Sexual Activity> Progression for Females>
Female Undetected CIN3 to
<Years in Years Undetected CIN1
Mortality Rate Regression CIN2
0 to 4> <Years in Years to CIN1 Rate Regression from
Deaths of Female Deaths of Female 5 to 9> Deaths of Female Regression (tau-16fbs21) Detected CIN2 to Aging of treated and
Children 0 to 4 Children 5 to 9 from Detected CIN1 infected CIN2 ()
Children 10 to 11 Undetected
Regression CIN3 CIN2 to
to CIN1 Rate Undetected <Regression CIN2 to
CIN1 CIN1 Rate
<Mortality Rates for (tau-16fbs31)
Regression from Regression CIN3 (tau-16fbs21)>
<Mortality Rates for Children 10 to 11 by
<Mortality Rates for Detected CIN3 to % CIN2 infected
Children 5 to 9 by Sex> Sex> to Infected Rate Treated and Infected CIN2 (ICIN2-16fb)
Infected after treatment
Children 0 to 4 by Sex> (tau-16fbs3)
(psi-s2)

Regression CIN2
to Infected Rate Progression from Detected CIN2 to
(tau-16fbs2) Treated & Infected CIN2

Regression from
Undetected CIN2 to Regression from Treated & Infected mu-ICIN2
Infected CIN2 to Undetected CIN2 death
Rate of waning natural
immunity (sigma-zf) Progression from <Female
Undetected CIN2 to Mortality Rate>
Detected CIN2 Cure Rate of CIN2
<Mean Time Until Age <Mean Time Until Age (cap-gamma s2)
Progression for Females> Progression for Females> <Detected CIN2 Rate Aging of
<Mean Time Until Age
(kappa-16fbis2)> detected CIN2 ()
Progression from Progression for Females> Detected CIN2 (DCIN2-16fb)
Immune to Susceptible
(f) Aging of
Aging of immune undetected CIN2 () Undetected CIN2 (CIN2-16fb)
Aging of <Mean Time Until Age
females () undetected CIN3 () Progression for Females> mu-DCIN2 <Female
Immune (Z-16fb) Undetected CIN3 (CIN3-16fb) Mortality Rate>
death

Progression from
muZ-fb Undetected CIN2 to
death Undetected CIN3
Undetected
mu-CIN3 death CIN2 Rate
<Female (pi-16fbs2)

Stratified by:
<Female Recovered from Mortality Rate>
Mortality Rate> Undetected Regression from
CIN3 Undetected CIN3 to
Undetected CIN2

<Regression Regression CIN3

CIN3 to Infected to CIN2 Rate Progression from
Rate (tau-16fbs32) Detected CIN2 to
(tau-16fbs3)> Treated & Cured TCINs
<% CIN2 infected after

Cumulative
<Fraction of CINs regressions treatment (psi-s2)>
clearing CIN that also clear mu-CIN2 death
infection (gamma bar-16f)>
<Female <Cure Rate of CIN2
Mortality Rate> (cap-gamma s2)>

Recovered from Undetected CIN2

<Fraction of CINs regressions <Regression CIN3 to
clearing CIN that also clear CIN1 Rate
infection (gamma bar-16f)> <Fraction of CINs regressions (tau-16fbs31)>
clearing CIN that also clear
infection (gamma bar-16f)>
Recovered from Detected CIN2

Recovered from Treated & Infected CIN2

<Rate of Recovery from
infection wtih HPV Regression from
<Regression CIN2 to
Infected Rate (gamma-16f)> Detected CIN3 to
Regression from
Detected CIN3 to (tau-16fbs2)> Detected CIN1
Progression from Undetected Detected CIN2 <Regression

17 age categories
CIN3 to Detected CIN3 CIN3 to CIN2
<Detected CIN3 Rate Rate
(tau-16fbs32)>
(kappa-16fbis3)>
<Mean Time Until Age
Progression
<Regression CIN3 to from Progression for Females>
Infected Rate Undetected
(tau-16fbs3)> CIN3 to
Undetected
Carcinoma s1
Aging of

Counts
detected CIN3 ()
Undetected <Mean Time Until Age
CIN3 Rate Progression for Females>
(pi-16fbs3)

Aging of treated and

Detected CIN3 (DCIN3-16fb) cured TCINs ()
Recovered from Detected CIN3 Treated and Cured TCINs (TCINs-16fb)
<Mean Time Until Age Progression from Detected Progression from Detected CIN3 to
Progression for Females> CIN3 to Undetected Treated & Cured TCINs
Carcinoma s1

Regression from
Treated & Infected
Aging of undetected % CIN3 infected
<Fraction of CINs regressions CIN3 to Undetected
carcinoma s1 () after treatment
clearing CIN that also clear CIN3
(psi-s3)
infection (gamma bar-16f)> mu-DCIN3
Progression from Cure Rate of CIN3
mu-CIS1 death (cap-gamma s3) mu-TCINs
<Female Detected CIN3 to
death death
Undetected Carcinoma s1 (CIS1-16fb) Mortality Rate> Treated & Infected
CIN3
Reoccurence of
CIN3 (theta-rs3) <Female

2 cervical screen. groups

Mortality Rate>
Regression from Treated & Infected
<Female Carcinoma s1 to Undetected Carcinoma s1
Mortality Rate>

Reoccurence
Progression from Undetected of CIS1
Carcinoma s1 to Undetected (theta-rs4)
Carcinoma s2 <Detected CIS1 Rate
(kappa-16fbis4)>
Progression from
Undetected Carcinoma s1
Undetected to Detected Carcinoma s1
CIS1 Rate
(pi-16fs4)

Recovered from
Treated & Infected Treated and Infected CIN3 (ICIN3-16fb)
CIN3

<Mean Time Until Age

Progression for Females> Aging of treated and
infected CIN3 ()

mu-DCIS1

3 sexual activity groups

<Rate of Recovery from death
infection wtih HPV <Mean Time Until Age
(gamma-16f)> Progression for Females> mu-ICIN3
death
<Female
Mortality Rate>

Aging of
detected
carinoma s1 ()
Treated and Infected Carcinoma s1 (ICIS1-16fb)
Detected Carcinoma s1 (DCIS1-16fb)

mu-ICIS1
Recovered from Treated death
& Infected Carcinoma s1 Progression from Detected
% CIS s1 infected Carcinoma s1 to Treated &
after treatment <Female
Infected Carcinoma s1
(psi-s4) Mortality Rate>
Progression from
Detected Carcinoma s1 to Cure Rate of CIS s1
Treated & Cured TCISs (cap-gamma s4)

Aging of treated and

Progression from Detected infected carcinoma s1()
Carcinoma s1 to Undetected
Carcinoma s2
<Mean Time Until Age
Progression for Females>
<Mean Time Until Age
<Undetected CIS1 Progression for Females>
Rate (pi-16fs4)>
Aging of treated

2 smoking statuses
<Rate of Recovery from
infection wtih HPV and cured TCISs ()
(gamma-16f)> Treated and Cured TCISs (TCISs-16fb)

mu-TCISs
death
<Female
Mortality Rate>
<Female
Mortality Rate>

Female
mu-ICIS2
Regression from Treated & Infected death
Carcinoma s2 to Undetected Carcinoma s2
Progression from
Detected Carcinoma s2 to % CIS s2 infected after
Reoccurence Cure Rate of CIS s2 Treated & Cured TCISs treatment (psi-s5)
of CIS2 (cap-gamma s5)
<Mean Time Until Age
(theta-rs5)
Progression for Females>

<Mean Time Until Age <Mean Time Until Age

Progression for Females> Progression for Females>
Aging of Aging of
undetected
carinoma s2 detected
() carcinoma s2 ()
Undetected Carcinoma s2 (CIS2-16fb) Aging of treated and
infected carcinoma s2 ()
Detected Carcinoma s2 (DCIS2-16fb)
Treated and Infected Carcinoma s2 (ICIS2-16fb)
Progression from Detected

2 sexes
Progression from Carcinoma s2 to Treated &
mu-CIS2 Infected Carcinoma s2
Undetected Carcinoma s2
death
to Detected Carcinoma s2
<Female
Mortality Rate> mu-DCIS2 death <% CIS s2 infected <Cure Rate of CIS s2
Progression from after treatment (cap-gamma s5)>
Undetected Carcinoma s2 <Detected CIS2 Rate <Female (psi-s5)>
to Undetected Local (kappa-16fbis5)> Mortality Rate>

Recovered from Treated

& Infected Carcinoma s2

Population
Undetected
CIS2 Rate
(pi-16fbs5)
<Female
Mortality Rate>
mu-CCR
death

Progression from Detected <Mean Time Until Age

Carcinoma s2 to Undetected Local Progression for Females>
mu-CCD death

<Mean Time Until Age <Female

Progression for Females> Aging of undetected Undetected Regional to Mortality Rate>
regional () Undetected Distant Rate Aging of
(pi-R) undetected distant
Undetected Regional (CCR-16fb)
Aging of () Undetected Distant (CCD-16fb)
undetected local ()
<Mean Time Until Age
Undetected Local (CCL-16fb) Progression for Females>
Progression from

Each visual stock represents

Undetected Regional to
Undetected Distant
Progression from
Undetected Local to
mu-CCL
Undetected Regional
death Undetected Regional to
Detected Regional Rate
(upsilon-R)
<Female
Mortality Rate> Undetected Local to
Undetected Regional
Rate (pi-L)
Progression from
Undetected Regional to
Progression from Detected Regional
Undetected Local to
Detected Local Progression from Cancer Death Total
Undetected Regional to Cancer Death by over Population
Cancer Death CervicalScreeningCategory
<Detected CIS1 Rate
(kappa-16fbis4)>
Undetected Local to Undetected Distant to
Cancer Death by
Detected Local Rate RCC death rate Cancer Death by Detected Distant Rate
SmokingStatus and
(upsilon-L) (chi-R) SmokingStatus (upsilon-D)
CervicalScreeningCategory
<Cancer Death by
AgeCategory SmokingStatus
and
CervicalScreeningCategory> Progression from
Undetected Distant to
<DCC death rate Progression from Detected Distant
(chi-D)> Undetected Distant
to Cancer Death
<Mean Time Until Age
LCC death rate
Progression for Females>
(chi-L)
Progression from
Undetected Local to
Aging of Cancer Death <Female
Mortality Rate>

408 distinct underlying stocks

detected local ()
Progression from
Detected Local (DCCL-16fb)
Detected Regional to Detected Regional (DCCR-16fb)
Cancer Death Cancer Death
mu-DCCR
Aging of detected death
mu-DCCL regional ()
death
Progession from
Detected Local to <RCC death rate
<Female Cancer Death (chi-R)>
Mortality Rate>

<Mean Time Until Age

<LCC death rate Progression for Females>
(chi-L)> Aging of detected
distant ()

DCC death rate (chi-D)

<Mean Time Until Age

Progression for Females>
Progression from
Aging of cancer Detected Distant to
survivors() Cancer Death

Cancer Survivors (SCC) Detected Distant (DCCD-16fb)

Progression from
Detected Local to Cancer
Survivors Progression from
Detected Distant to
Cancer Survivors

mu-SCC
death Detected Distant
Cancer Survivor mu-DCCD
Detected Local Cancer Rate (omega-D)
Survivor Rate death
(omega-L) <Female
Mortality Rate>
<Female
Mortality Rate>

Progression from
Detected Regional to Detected
Cancer Survivors Regional
Cancer
Survivors Rate
(omega-R)
Example Mixing Preferences
 Sources for Parameter Estimates
Surveillance data
Controlled trials
Outbreak data
Clinical reports data
Intervention
outcomes studies
Calibration to
historic data
Expert judgement Anderson & May
Systematic reviews
 Introduction of Parameter Estimates
Frequently System Dynamics models will provide much more detail on networks of
factors shaping these rates, but ultimately there will be constants requiring specification

<Annual Likelihood of
Annual Likelihood of Non-Diabetes Mortality for
Becoming Diabetic Asymptomatic Population>

<Annual at Risk
Births> undx uncomplicated
dying other causes

Developing
Diabetes Undx Prediabetic
Being Born Non
Popn
Obese Being Born At
Risk
Undx Prediabetics
Recovering
<Annual Not at Annual Likelihood of
Non Obese Becoming Obese Obese General
Risk Births> General Population Annual Likelihood of Diagnosis of
Population
Becoming Obese Undx Prediabetic prediabetics
Recovery
Annualized Probability
Density of prediabetic
recongnition
Annualized Mortality Dx Prediabetics
Rate for obese Recovering
population
Obese Mortality Dx Prediabetic Popn

Annual Likelihood of Dx
Prediabetic Recovery

Annual Mortality Rate for

Non-Obese
non obese population
Mortality
dx uncomplicated
dying otehr causes

Annual Likelihood of
Non-Diabetes Mortality for
Scenarios for Understanding
How Does X affect System
Single Model Matches Many Data Sources

one of
 Example Aggregate Model Structure

Population Size

Per Contact Risk of

Immigration Rate Contacts per Infection Fractional
Susceptible Average Duration of
Prevalence Infectiousness

Susceptible Infective Recovered

Incidence Recovery
Immigration
 Mathematical Notation

Population Size
N

Per Contact Risk of
Contacts per Infection
c Fractional

Susceptible Mean Time with
Prevalence
M
Immigration Rate Disease

S
Susceptible
Absolute
I
Prevalence
R
Recovered
Immigration of Incidence Recovery
Susceptibles
Underlying I
S M c S
(Ordinary) N
Differential
Equations I I
I c S
N
I
R

 Model Mathematical Analysis
System Linearization (Jacobian)

I
S c S R 0 State space diagram (reasoning about
Fixed-Point Criteria N
I I many scenarios at once)
I c S 0

N
I
h
I
R R 0
I

h
Eigenvalues (e.g. for stability analysis around fixed-point)
 Some Uses of Formal Approaches
Explaining observed behavior patterns
Identifying possible behavior modes over a
wide variety of possible scenarios (e.g. via
eigenspace & phase plane analysis)
Identifying how behavior depends on
parameters (stability, location of equilibria)
Creating self-correcting models (via
control theory)
Formal calibration methods
Feedbacks Driving Infectious Disease
Dynamics
Susceptibles
+
-
+
Contacts
Suscepti between New Infections
Infectilveess and
b

+ +

Infectives

-
+

New Recoveries
 Example Dynamics of SIR Model (No Births or
Deaths)
SIR Example
2,000 people
600 people
10,000 people

1,500 people
450 people
9,500 people

1,000 people
300 people
9,000 people

500 people
150 people
8,500 people

0 people
0 people
8,000 people
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200
Time (days)

Susceptible Population S : SIR example people

Infectious Population I : SIR example people
Recovered Population R : SIR example people
 Shifting Feedback Dominance
SIR Example
2,000 people
600 people
10,000 people Susceptibles
+
-
+
Contacts
Suscepti between New Infections
Infectiblveess and
1,500 people +
+
450 people
Infectives
9,500 people
-
+

1,000 people New Recoveries

300 people
Susceptibles
9,000 people +
Susceptibles
+
- -
+ +
Contacts
Suscepti between New Infections Contacts between
Infectiblveess and Suscepti New Infections
Infectiblveess and
500 people + + + +

150 people Infectives

Infectives
8,500 people
-
+
-
+
New Recoveries
New Recoveries
0 people
0 people
8,000 people
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200
Time (days)

Susceptible Population S : SIR example people

Infectious Population I : SIR example people
Recovered Population R : SIR example people
 Dynamic Uncertainty:
Stochastic Processes
(Stochastic Differential Equations)
Baseline
50% 60% 70% 80% 90% 95% 98% 100%
Average Variable Cost per Cubic Meter
0.6

0.45

0.3

0.15

0
0 1457 2914 4371 5828
 Stakeholder Engagement with
Created Models

Team Meetings

Mabry, 2009, Simulating the Dynamics of Cardiovascular

Health and Related Risk Factors
 Varied Applications of
System Dynamics
Aggregate population-level models
Stratified models
Models of an individual
Models of interactions of two or more
individuals
Stochastic models
 Key Take-Home Messages on SD
Focuses on feedbacks as the fundamental
shapers of dynamics
Models are consciously specific to purpose
Includes both qualitative & quantitative
components
Offers strong stakeholder focus
Group model building
Stakeholder learning laboratories
Can be applied at diverse levels of
granularity
Models admit to formal reasoning & analysis
Department of Computer Science
System Science Methodologies: Highly
Complementary
No one system science methodology offers
a replacement for the others
Significant synergies can be secured by
using combinations of methodologies to
address the same problem
As cross-checks on understanding where two
or more can be applied
Exploiting competitive advantages
 Multi-Framework Modeling
We have found the use of multiple
frameworks most effective
Co-evolving multiple models for
Cross-validation
Asking different sorts of questions
Within a single model (cf Multi-scale modelling)
Critical that dynamic models leverage with
non-dynamic modeling tools
Decision trees
Game theory
Biostatistical analyses
Multiple Modeling Types

System
Dynamics

Deriving calibrated parameter estimates for low-level model

Focusing AB exploration Agent-
Inspiring key initial structure of agent-based models Based
Diagramming out high-level drivers of behaviour
Description of continuous individual-level evolution Modeling

Social
Network
Analysis
Network Embedded Individuals
Virion Production Rate if Virion Production Rate Mean of Viral Load
1 Person Non Quantized Infection Per Contact Virions Rate of Neighbors
Mean Viral Load
<Population Size> Virus Load
Virion Production Virion Clearance
From Infected Cells
Likelihood Density of Mean Virion
Uninfected Cell
Infection by Single Virion Lifetime
Replentishment Rate
Per Infected CellVirion
Uninfected Infected Production Rate
Cells Cells infected cell death
Uninfected Cell New Cell
Replentishment Infections by CTLs rate which infected cells
Uninfected Cell Infected Cell
Death are killed by CTLs
death

<Population Size>
Mean Infected
Mean Uninfected Mean Uninfected Cells
<Population Size> Cell Lifetime
Cells Mean Infected Cell
Lifetime

CTL CTLs Mean CTL

responsiveness immune response to CTL turnover lifespan
infected cells
 Multiple Modeling Types

System
Dynamics

Cross-validating SD aggregation
Giving insight into feedbacks to depict
Evaluating dynamic importance of
Agent-
stratifying to capture heterogeneities Based
Modeling

Social
Network
Analysis