Treffer: Fine-grained synthetic population generation and agent-based models for COVID-19 in Malta

This study presents the development of a fine-grained COVID-19 agent-based model (Agent-based Model (ABM)) specifically designed for Malta, leveraging a synthetic population that captures the country’s demographic and tourism characteristics. The research is structured into three phases. In Phase 1, the SynthPops framework is extended to generate a statistically accurate synthetic population, enriched with additional attributes such as employment, education, BMI, and long-term illnesses. A detailed tourism model is also integrated to reflect Malta’s unique visitor dynamics. Phase 2 focuses on implementing the ABM, which incorporates detailed daily itineraries, a contact network, and virus transmission dynamics. Transmission is influenced by factors such as individual sociability, contact duration, and public health interventions. The model is used to simulate multiple intervention scenarios, producing epidemiological outcomes that align closely with the input parameters and provide actionable insights. In Phase 3, the study evaluates four computational strategies to optimise execution time and scalability: single-node multiprocessing and three distributed approaches using Dask Distributed. Among these, the actor-based strategy demonstrates the best performance, achieving up to a 13-fold speed-up in specific tasks and scaling effectively with population size. Testing on a high-performance machine reveals that the model performs well for Malta’s population size, with distributed setups showing potential for larger populations. This research provides a robust and scalable framework for simulating COVID-19 dynamics in islands such as Malta, offering valuable insights for public health decision-making and highlighting computational strategies for efficient large-scale simulations. ; peer-reviewed