Microsimulation Calibration Integrating Synthetic Population Generation and Complex Interaction Clusters to Evaluate COVID-19 Spread

CG Gómez-Marín, JD Mosquera-Tobón… - Handbook on Decision …, 2022 - Springer
Handbook on Decision Making: Volume 3: Trends and Challenges in Intelligent …, 2022Springer
Agent-based microsimulation is a modeling technique used mostly in social sciences;
nevertheless, its bottom-up approach to describe a system is helpful for engineering fields
as well. Such modeling technique starts by generating a synthetic population to set the units
of analysis in the simulation. Some authors have used large synthetic populations of entire
cities for different purposes such as solving transportation problems, conducting spatial
analysis, and analyzing health care systems. However, generating synthetic populations is …
Abstract
Agent-based microsimulation is a modeling technique used mostly in social sciences; nevertheless, its bottom-up approach to describe a system is helpful for engineering fields as well. Such modeling technique starts by generating a synthetic population to set the units of analysis in the simulation. Some authors have used large synthetic populations of entire cities for different purposes such as solving transportation problems, conducting spatial analysis, and analyzing health care systems. However, generating synthetic populations is still a difficult task, especially for microsimulation models that evaluate COVID-19 spread. Therefore, this chapter proposes an algorithm that generates a synthetic population to evaluate COVID-19 spread in a microsimulation scenario. Each individual in the population is characterized by specific features, such as age, gender, comorbidities, work activities, and school activities. Moreover, the population of individuals is grouped into four clusters: home, work, school, and shopping. In each cluster, the social interaction is stablished using scale-free fully-connected networks. This proposal was validated using data on the second largest city in Colombia, i.e., Medellín, achieving quite promising results (which are close to actual microdata on the population of said city) and a total average error of 4.53%. Future studies could add some new clusters, such as transportation and geographic location.
Springer
以上显示的是最相近的搜索结果。 查看全部搜索结果