Research
We combine mechanistic modelling, statistical inference and machine learning to study infectious disease, from the dynamics inside a single host to transmission across whole populations.
Fig. 2 Interactive model
Change the reproduction number, the infectious period, vaccine coverage and the strength of seasonality, and watch the epidemic respond. This is the kind of model our work starts from.
Average number of people one case infects in a fully susceptible population.
The gold mark shows the herd immunity threshold, given by 1 − 1/R0.
β(t) = β0[1 + a cos(2πt/365)], the kind of time-varying rate studied in our work on influenza.
Fig. 2. A deterministic SIR model with vaccination before the outbreak and sinusoidal seasonal forcing, solved in your browser with a fourth-order Runge-Kutta scheme. Shares are of the whole population; one person in ten thousand is infectious on day 0.
01 Research areas
Each area has its own methods, but most projects draw on several of them at once.
Disease processes unfold on very different timescales, from rapid molecular interactions to long-term population dynamics. We develop mathematical frameworks, including singular perturbation methods, that connect molecular, cellular, individual and population-level dynamics.
Models of immune response dynamics that show how pathogens interact with host immunity, covering antibody responses, T-cell dynamics and cytokine signalling networks.
How geographical and socio-economic factors shape transmission and outcomes. We combine spatial statistics with epidemiological models to identify risk factors and anticipate where disease will spread.
Statistical and machine-learning methods for epidemiological data, including Bayesian inference, time-series analysis, deep neural networks and random forests.
Disease-informed neural networks that build epidemiological principles into deep learning architectures. These hybrid models aim to make forecasts of outbreaks and of disease progression more accurate.
Mathematical models for public health and policy, including the analysis of nonlinear dynamical systems, differential equation models of disease and the optimization of vaccination strategies.
02 Approach
Results in one area feed into the others. Hover over or tap any part of the diagram to see its connections.
03 Projects
A selection of the work under way in the lab.
Active
Modelling how dengue virus interacts with the immune system, to understand what makes disease severe and to inform better treatment.
Active
Mapping the geographical distribution of mpox and the socio-environmental indicators associated with risk in Ontario.
Active
Neural networks that predict the spread of avian influenza from clinical, genomic, environmental and demographic data.
Active
Models of vaccination strategies that optimize coverage and long-term effectiveness across populations.
Active
Deep neural networks and random forests that find meaningful patterns in complex epidemiological data.
04 Applications
05 Collaboration
