Bayesian-based predictions of COVID-19 evolution in Texas using multispecies mixture-theoretic continuum models

Area of Science:

• Epidemiology
• Mathematical Modeling
• Computational Biology

Background:

• COVID-19 spread necessitates accurate predictive models.
• Real-time data integration is crucial for public health.
• Understanding disease dynamics requires robust mathematical frameworks.

Purpose of the Study:

• To develop and calibrate a mixture-theoretic continuum model for COVID-19 transmission in Texas.
• To perform real-time model calibration and uncertainty quantification using Bayesian learning.
• To predict COVID-19 deceased and infected cases and validate model performance.

Main Methods:

• Utilized a system of coupled partial differential reaction-diffusion equations.
• Employed a Bayesian learning approach (Occam Plausibility Algorithm - OPAL) for model calibration and prediction.
• Incorporated real-time COVID-19 data for parameter estimation and uncertainty quantification.

Main Results:

• The model indicated lower mortality rates in Texas compared to existing literature.
• A prediction of 7003 deceased cases by September 1, 2020, was made with a 95% confidence interval of 6802-7204.
• The model was validated for total deceased cases but found to be invalid for total infected cases.

Conclusions:

• The developed model provides valuable insights into COVID-19 dynamics in Texas, particularly regarding mortality.
• Model limitations were identified, especially in predicting total infected cases, suggesting areas for future refinement.
• Further improvements to the model are necessary for enhanced predictive accuracy and broader applicability.