Inverse Mathematical Modelling in Epidemiology

Start date: 8 December 2020
End date: 7 December 2023
Value: £11,200 RG.MATH.122518
Primary investigator: Professor Daniel Lesnic
External co-investigators: O. Krivorotko

As of May 2020, the World Health Organization estimated that over 5 million COVID-19 cases have occurred globally, out of which over 300,000 people died. Moreover, new species of viruses emerge at a rate of 3-4 per year. While direct modelling of population dynamics of infectious diseases is well-understood, in reality, many properties of biological processes are in fact unknown and their direct measurement is demanding in terms of experimental time and resources, possibly very inaccurate or even impossible. In such a situation, the project proposed an inverse modelling approach based on parameter identification/estimation from additional information/measurement.

This new collaboration was aimed at developing a suite of mathematically deterministic and statistically stochastic models to support each other and further develop the reliability of predictions in the field of epidemiology. Due to the COVID-19 situation the applicability of this exchange project was very timely.

Impact

Inverse modelling is crucial for calibrating a mathematical model and for controlling/identifying the model coupling parameters. Designing new globally convergent iterative methods of minimization for parameter identification in epidemiology represented a reliable step forward in the analysis. When applied to the spread of infectious diseases such as COVID-19, the inverse problem approach may enable public health scientists and epidemiologists in the UK (and indeed elsewhere) to predict the time interval between epidemics. The rigour of this inference approach based on parameter identification/estimation, as opposed to direct inaccurate measurement, may add confidence to UK policy makers and the NHS.

The success of the research on parameter identification for transmitted-disease modelling has initiated and propelled the individual participants in this project into the new field of mathematical epidemiology. The two-way knowledge transfer was further realised through the high-impact joint paper that was elaborated.

Publications and outputs

Krivorotko, O., Sosnovskaia, M., Vashchenko, I., Kerr, C. and Lesnic, D. (2022) Agent-based modeling of COVID-19 outbreaks from New York state and UK: Parameter identification algorithm, Infectious Disease Modelling, Vol. 7, No. 1, pp. 30-44.