Sensitivity, Estimability and Parameter Identification of a Mathematical
Covid-19 Epidemic Model
- Ilias Bouchkira
Abstract
In this work, a mathematical Covid-19 epidemic transmission network
model is investigated for the study of the virus spread dynamics. The
purpose is to present a sensitivity-based estimability analysis as well
as an accurate parameter identification approaches for reliable
mathematical modeling. The most sensitive parameters of the model are
identified using a local sensitivity approach, these sensitivities are
then used within an orthogonalization algorithm to assess the
estimability of the unknown parameters from available data. A database
of newly reported infected and recovered people in China is used. The
most estimable model parameters are identified, their accuracy is
assessed by computing confidence intervals and their numerical values
are also used to compare the model predictions to real data. The Pearson
Product-Moment coefficient is computed. Its high values show the
accuracy of the new model parameterization and the importance of the
used sensitivity-based estimability approach.