Abstract:
Globally, the COVID-19 pandemic has claimed millions of lives. In this study, we
develop a mathematical model to investigate the impact of human behavior on the dynamics of
COVID-19 infection in South Africa. Specifically, our model examined the effects of positive versus
negative human behavior. We parameterize the model using data from the COVID-19 fifth wave of
Gauteng province, South Africa, from May 01, 2022, to July 23, 2022. To forecast new cases of
COVID-19 infections, we compared three forecasting methods: exponential smoothing (ETS), long
short-term memory (LSTM), and gated recurrent units (GRUs), using the dataset. Results from the time
series analysis showed that the LSTM model has better performance and is well-suited for predicting
the dynamics of COVID-19 compared to the other models. Sensitivity analysis and numerical
simulations were also performed, revealing that noncompliant infected individuals contribute more
to new infections than those who comply. It is envisaged that the insights from this work can better
inform public health policy and enable better projections of disease spread.
