Epidemic Modeling in NetLogo

1
Epidemic Modeling in NetLogo

• Brendan Greenley Pd. 3

2
Purpose

• Create a simple yet realistic model of an
  epidemic
• Figure out how manipulating variables changes the
  behavior of an epidemic.

3
Goals

• Identify how changes in variables such as death
  rate, virus duration, and time until quarantine
  affect the behavior/duration of an epidemic
• Allow for simple tweaking of variable through
  sliders
• Implement System Dynamics (SD)

4
Program Run

• Run time possibilities
• Short term (one epidemic)
• Long term (reoccurring epidemics)
• Key End Results
• Death count
• Length of outbreak
• Time with maximum of infections
• Results shown graphically (real-time)

5
Similar Projects

• Numerous epidemic models
• None implement System Dynamics
• More complex
• Parametrics
• Differentials
• Basic assumptions frequently the same
• Infections can only spread from sick -gt
  unaffected
• Those who survive gain immunity (unless virus
  mutates)
• Those infected but quarantined cannot spread
  virus

6
NetLogo

• Programming language (Northwestern)
• More popularly used in multi-agent based
  programming
• My use System Dynamics
• Crossplatform support
• Windows, Nix, Mac
• Free!

7
Procedure

• Add stock/flows in smaller groups
• Check to see if smaller, simpler linkages work
  properly by tracking stock populations in test
  runs
• Attempt to link smaller linkages into one greater
  system
• Check with a test run and repeat until project
  has one, huge working system of flows.

8
Time Line

• First quarter
• Learn how to use NetLogo
• Experiment with non-SD procedures
• Second
• Successfully create a basic model that
  encompasses unaffected, infected, quarantined,
  and immune stocks
• Third
• Add more variables and flows to model
• Attempt to have epidemics repeat over a longer
  period of time (centuries), with different
  variations and mutations expressed by a change in
  variables
• Fourth
• Focus on data collection and make conclusions
  from data, look at derivatives of graphs, etc.

9
Problems

• Changing rates over time
• As awareness of disease increases, so should
  quarantine rate
• Using flows realistically
• Balancing population shifts with eachother
• Combining System Dynamics and non-SD components
  can be difficult
• Ticks
• Should a tick represent a day? An hour?

10
Post-development Possibilities

• What if I finish early
• Try to create same model in NetLogo, but without
  using System Dynamics
• Extend my epidemic model so it can be used to
  model long term diseases like HIV/AIDS
• Agents to represent the populations and have them
  shown on a GUI

11
Results

• Due to natural immunities, killing off an entire
  population in one epidemic is difficult.
• Viruses that are too deadly are poor diseases,
  they quickly die off.
• Quarantining is a very effective measure to slow
  the infection rate.
• My SD model yields smoother curves than my non-SD
  model, (though there are slightly different
  algorithms/values used)

12
System Dynamics Model
13
Non-System Dynamics Model
14
Plan Changes

• I shifted away from System Dynamics, but came
  back to it
• SD environment yields less mistakes fewer
  chances for typos/forgetting to update variables
  than non-SD NetLogo
• I initially was going to do agent based modeling,
  but that is difficult to do with System Dynamics.