The COVID-19 pandemic has evidenced that reliable model-based epidemiological predictions have remained an open challenge, and this concerns in particular the identification of model parameters that may change throughout the course of the pandemic. This aging characteristic of an epidemic is our present focus, by example of predicting fatality trends from infection histories. Regarding the challenge as a mechanobiological problem, we employ a hereditary mechanics-rooted Boltzmann-Volterra-type integro-differential equation, so that the fatality rate is obtained as a time integral over the rate of confirmed cases (i.e., the so-called incidence), multiplied with a kernel depending on evolving, i.e., time-dependent, fatality fractions and time lags between infection and death. This novel convolution approach, including its degeneration to an aging infection-to-death-rate delay rule, is superior to the traditional kinetics approach in as many as 93% of the tested cases associated with 228 countries, territories, and US states. The corresponding country-, territory-, and US state-specific fatality fractions appear as exponentially decaying quantities with characteristic decay times ranging from around 100 days to several years; with a world median of some 480 days, and with 100 to 200 days being typical for Western Europe and several Eastern US states. These parameters show periods of fair stability over one to several hundred days, indicating midterm prediction capabilities of our novel approach.
en
Research Areas:
Mathematical and Algorithmic Foundations: 10% Modeling and Simulation: 90%
