Agent-based simulations for optimized prevention of the spread of SARS-CoV-2 in nursing homes

Complexity Science Hub Vienna     |     Medical University Vienna

Jana Lasser     |     lasser@csh.ac.at     |     @janalasser


slides available at
https://janalasser.at/talks/nursing_home_COVID_talk/

motivation

  • High mortality
  • Close living conditions in homes → easy spread
  • Balance between quality of life and safety

aim

Use agent based simulations

in a model of an Austrian nursing home

to find the best prevention strategy.

agent states

Collaboration with Caritas to model Austrian nursing homes
floor plan and interactions

contact network

SEIR-dynamics

test types

agents are quarantined, the outbreak dies off eventually, without infecting all residents
→ adjust transmission risk and contact weights to reproduce outbreak data
employees were index case in 8/8 recorded outbreaks
employee index case:    R0  =  2.61 ± 1.33
resident index case:    R0  =  4.82 ± 2.01


tests per person per day

Different test technologies

Different turnover times

technical notes

  • existing stable library
  • easy to extend
  • out-of-the-box functionality 
    
						  				self.datacollector.collect()
  									
    
						  				self.schedule.step()
Mesa: Agent-based modeling in Python