Differential Equations Seminar: Undergraduate Researcher

Abstract: Epidemic forecasts often require the compilation of highly stochastic, error-prone data. For tracking epidemics in real-time, there must be reliable predictions of the spatial-temporal spread in a critical window of time in order to contain an outbreak. In my research, I applied the data assimilation method, Optimal Interpolation, to track the spread of an SEIRD (Susceptible-Exposed-Infected-Recovered-Dead) model, a compartmental model of an infectious disease. The model incorporates synthetic disease incidence data of an Ebola epidemic for three cities in Nigeria (Abuja, Gombe, and Makurdi). Via the R programming language, Optimal Interpolation assimilates the incoming data from the SEIRD epidemic model to produce predictions of the spread in weekly increments. The accuracy of the predictive model is then compared to the original, with both side-by-side forecast images comparisons and time-series. Finally, we also apply the Optimal Interpolation data assimilation method to forecast the recent Coronavirus (COVID-19) pandemic. The model incorporates actual disease incidence data from government situation reports for the 20 Nigerian states that currently have confirmed cases.