Time-variant reliability-based prediction of COVID-19 spread using extended SEIVR model and Monte Carlo sampling

Area of Science:

• Epidemiology
• Public Health
• Biostatistics

Background:

• Coronavirus disease 2019 (COVID-19) has presented significant public health challenges globally.
• Accurate prediction of disease spread is crucial for effective intervention strategies.
• Existing models often require enhancements to incorporate factors like vaccination and quarantine.

Purpose of the Study:

• To develop and apply a probabilistic method for predicting the COVID-19 spreading profile in the United States.
• To utilize a time-variant reliability analysis framework for disease forecasting.
• To quantify the uncertainty associated with COVID-19 case predictions.

Main Methods:

• An extended susceptible-exposed-infected-vaccinated-recovered (SEIVR) deterministic model was formulated.
• The SEIVR model was integrated into a reliability framework with Monte Carlo sampling.
• Time-variant reliability analysis was employed to calculate exceedance probabilities (EPs) over time.

Main Results:

• The probabilistic model successfully predicted time-series data for infected, recovered, and deceased COVID-19 cases in the US.
• Exceedance probabilities were calculated for maximum infected, total recovered, and total fatal cases.
• 3D probability graphs illustrated the interplay between case numbers, exceedance probability, and time.

Conclusions:

• The study provides a robust probabilistic approach for forecasting epidemic dynamics.
• By the end of 2021, projections indicated approximately 0.8 million active COVID-19 cases in the US.
• The model also projected cumulative recoveries of 41.3 million and 0.6 million fatalities by the end of 2021.