Continuous-discrete GeoSEIR(D) model for modelling and analysis of geo spread COVID-19

Humanity faces various types of viral infections, such as COVID-19, annually. In this paper, we propose a Geospatial SEIR(D) model based on a multi-agent approach with continuous-discrete states. This model accounts for key parameters of viral infections, daily human activities, and geodata. Our developed algorithms enable the simulation of statistical parameters such as the number of infected, recovered, deceased, and susceptible individuals, along with the spatial distribution of the pandemic on a geographical map. The model was validated by simulating the COVID-19 spread in Lviv, Ukraine. Several preventive strategies were analyzed: implementing a 50 % reduction in infection probability through mask mandates delayed the peak to 150 days with a 25 % reduction in the maximum number of patients, while a 75 % reduction delayed the peak to 240 days with a 60 % reduction in the maximum number of patients. Prohibiting public transport and public places resulted in the epidemic peaking on day 165 with 2854 patients, significantly reducing the spread rate compared to the base model. Simulating 50 %, 75 %, and 100 % vaccination rates showed a reduction in the peak number of infections by 34 %, 57 %, and 94 %, respectively, also extending the duration of the epidemic. Enforcing weekend quarantine delayed the epidemic onset by one month but had minimal impact on the overall number of infections and duration. Combining mask mandates, transport restrictions, and vaccination led to the most effective mitigation, with the average number of sick agents around 8 and never exceeding 15 over four years. This comprehensive approach highlights the effectiveness of combining various preventive measures to control the spread of viral infections. The proposed model provides a valuable tool for policymakers to evaluate and implement effective strategies against pandemics.