Visual Analytics of Contact Tracing Policy Simulations During an Emergency Response

Abstract

Epidemiologists use individual-based models to (a) simulate disease spread over dynamic contact networks and (b) to investigate strategies to control the outbreak. These model simulations generate complex 'infection maps' of time-varying transmission trees and patterns of spread. Conventional statistical analysis of outputs offers only limited interpretation. This paper presents a novel visual analytics approach for the inspection of infection maps along with their associated metadata, developed collaboratively over 16 months in an evolving emergency response situation. We introduce the concept of representative trees that summarize the many components of a time-varying infection map while preserving the epidemiological characteristics of each individual transmission tree. We also present interactive visualization techniques for the quick assessment of different control policies. Through a series of case studies and a qualitative evaluation by epidemiologists, we demonstrate how our visualizations can help improve the development of epidemiological models and help interpret complex transmission patterns.