Forecast modelling for combination disease comorbidity
- Researcher: Rahil Sachak-Patwa
- Academic Supervisors: Helen Byrne and Robin Thompson
- Industrial Supervisors: Claude Schmit, Erin Lafferty
Background
The ability to accurately predict the course of an epidemic through a population, and the evolution of a disease in an individual, is imperative when making decisions about how to control the spread of a virus and treating a patient. In this thesis, we formulate mathematical models which can be used to describe and forecast viral disease dynamics.
Figure 1: Caption needed
In this project, we examine how data and modelling capabilities enable forecasting of dangerous disease combinations in low- and middle-income countries, where the burden of combination diseases is highest, but data are currently limited. We investigate a variety of data that can be used to inform epidemiological models, improve forecasts, and reduce disease incidence.
Outcomes
In our first project we model influenza epidemics. Previous exposure to influenza viruses confers partial cross-immunity against future infections with related strains. However, this effect is not always accounted for explicitly in mathematical models used for forecasting during influenza outbreaks. We consider two infectious disease outbreak forecasting models and show that in in scenario that we consider of a future epidemic occurring in Japan, a simple ‘1-group’ model only produces accurate outbreak forecasts once the peak of the epidemic has passed. As a result, we conclude that a more epidemiologically realistic ‘2-group’ model should be used to generate accurate forecasts.
Figure 2: Caption needed
In our second project, we consider describing and forecasting within-host influenza infections. Using synthetic data, we show that a simple a mathematical model consisting of only two variables can describe the viral dynamics in groups of hosts with different immunological characteristics. Furthermore, the same model can be used to accurately make forecasts in real-time of disease severity and can be used to classify patients as being comorbid or not based on the model parameters.
Publications
- Sachak-Patwa, R., Byrne, H. M., Thompson, R.N. Accounting for cross-immunity can improve forecast accuracy during influenza epidemics. Epidemics (2020): 100432. Link