28 Models Later: Model Competition and the Zombie Apocalypse

Area of Science:

• Mathematical modeling
• Epidemiology
• Game theory

Background:

• Humans versus Zombies (HvZ) is a popular campus game simulating zombie infection.
• Traditional SIR models with mass-action dynamics inadequately represent HvZ transmission.
• Understanding HvZ dynamics can offer insights into real-world social behaviors and disease spread.

Purpose of the Study:

• To identify key mechanisms driving the dynamics of the Humans versus Zombies (HvZ) game.
• To compare various transmission models using data from HvZ games.
• To validate model predictions with subsequent game data.

Main Methods:

• Developed a suite of mathematical models with diverse transmission mechanisms.
• Employed model competition, using Bayesian Information Criterion (BIC) for model selection.
• Utilized bootstrapping to assess mechanism significance and model confidence.
• Fit models to Fall 2011 data and tested predictions with Fall 2012 data.

Main Results:

• Humans exhibit defensive clustering behavior.
• Zombies demonstrate a tendency to hunt in coordinated groups.
• Individual zombie hunting proficiency varies.
• A portion of human players withdraw from the game over time.

Conclusions:

• The study identified specific behavioral mechanisms crucial to HvZ game dynamics.
• Model competition effectively distinguished between different transmission hypotheses.
• Findings suggest that HvZ dynamics are influenced by social and behavioral factors beyond simple infection rates.