Speaker
Description
COVID-19 affects diabetic and non-diabetic individuals differently, requiring models that capture heterogeneous disease dynamics. We propose a nonlinear compartmental model coupled with a Disease-Informed Neural Network (DINN) to study the spread of COVID-19 in these two risk groups. The model divides the population into susceptible, infected diabetic, and infected non-diabetic classes, with transitions governed by biologically meaningful parameters. The system is embedded into a neural network framework, where the state variables are approximated as functions of time while the governing differential equations are enforced through residual loss terms. The total training loss combines data regression errors with model-based constraints, ensuring both accuracy and consistency with disease dynamics. The results indicate that DINNs provide a robust and flexible tool for modeling complex epidemic systems and offer a promising framework for data-driven analysis of COVID-19 in vulnerable populations.
Bibliography
\bibitem{mona1} Anand, Monalisa, Danumjaya, Palla and Rao, Ponnada Raja Sekhara, Influence of incubation delays on COVID-19 transmission in diabetic and non-diabetic populations – an endemic prevalence case, Comput. Math. Biophys., 11(1), (2023) 20230115.
\bibitem{mona} M. Anand, P. Danumjaya, P. R. S. Rao, A nonlinear mathematical model on the Covid-19 transmission pattern among diabetic and non-diabetic population, Math. Comp. Simul., 210, 346-369 (2023).
