OpenABM-Covid19

OpenABM-Covid19 simulates SARS-CoV-2 transmission and evaluates non-pharmaceutical interventions, including manual and digital contact tracing, using an agent-based modelling framework.

Key Features:

• Agent-based modeling: Individual-based simulation of transmission dynamics for COVID-19 (SARS-CoV-2).
• Age stratification: Detailed age-structured population representation to capture age-dependent contact and disease patterns.
• Realistic social networks: Explicit modelling of social contacts to reflect heterogeneous interaction patterns.
• Contact tracing: Simulation of both manual and digital contact tracing within the epidemic model.
• Population scale and performance: Capable of simulating up to 1 million individuals with a reported runtime around one simulated day per second.
• Parameterization and calibration: Default parameterization uses UK demographics and calibration to UK epidemic data, with flexibility for re-parameterization to other settings.
• Intervention packages: Ability to simulate dynamic combinations of non-pharmaceutical interventions.
• Computational experimentation: Supports extensive parameter sweeps and formal statistical model-based inference for scenario analysis.
• Modularity: Model implemented with modular components to support testing and adaptation.

Scientific Applications:

• Evaluation of NPIs: Quantitative assessment of manual and digital contact tracing and other non-pharmaceutical interventions on SARS-CoV-2 spread.
• Intervention comparison: Comparative analysis of dynamic packages of interventions to assess relative impacts on epidemic suppression.
• Scenario analysis: Exploration of epidemic trajectories under varied demographic and intervention parameterizations.
• Parameter inference: Support for formal statistical model-based inference to estimate model parameters from epidemic data.
• Demographic impact studies: Investigation of age-dependent transmission and intervention effects using age-stratified populations.

Methodology:

Agent-based simulation with age-stratified populations and realistic social networks; parameterized to UK demographics and calibrated to UK epidemic data; explicit simulation of manual and digital contact tracing; scalable to ~1 million individuals with reported performance of ~1 simulated day per second; supports parameter sweeps and formal statistical model-based inference.