The COVID-19 pandemic brought infectious diseases to the fore of the world’s attention, demonstrating the massive and highly dynamic impact that a microorganism can have. This dynamic nature is fundamental to why infectious diseases are so difficult to control—infections spread exponentially, population immunity changes constantly, and interventions achieve impact nonlinearly. These hallmarks of infectious disease biology mean that mathematical models that capture dynamic phenomena are indispensable tools for combatting infectious diseases and improving the human condition.
While the appropriateness of mathematical models for addressing infectious diseases is clear, there are numerous outstanding challenges that obstruct progress towards using them to better control infectious diseases. One challenge is the fact that disease surveillance data often cannot be taken at face value and belie the true burden of disease in a population. A second challenge is that knowledge about interventions from clinical trials does not always translate into straightforward predictions of intervention impact when deployed on larger scales or in contexts that differ from that of a trial. This first challenge is one of understanding infectious disease burden and the second is one of predicting how infectious diseases will respond to interventions. Our research is organized around addressing these two challenges.
Mechanistic models are the core of our approach. Whereas statistical models capture phenomenological associations among variables (e.g., Y depends linearly on X), the structure and parameters of mechanistic models are derived from biological principles and assumptions (e.g., the rate at which susceptible individuals become infected depends on the rate at which they come into contact with infectious individuals and the probability that transmission occurs given contact). To address the first challenge of understanding infectious disease burden, we exploit the fact that models can synthesize information about multiple components of a system and generate new understanding of biological quantities that often cannot be observed directly. To address the second challenge of predicting how infectious diseases will respond to interventions, we leverage the fact that mechanistic models are equipped to explain differences in disease burden in space, time, and in other respects, which enables predictions to be made about intervention impacts in contexts that cannot be observed directly. Across the diverse set of projects in our lab, we employ mechanistic models with a variety of forms (ranging from simple equation-based models to complex agent-based models) and complement them with statistical approaches as needed. We frequently collaborate with empiricists to enrich the relevance and impact of our work.
While the appropriateness of mathematical models for addressing infectious diseases is clear, there are numerous outstanding challenges that obstruct progress towards using them to better control infectious diseases. One challenge is the fact that disease surveillance data often cannot be taken at face value and belie the true burden of disease in a population. A second challenge is that knowledge about interventions from clinical trials does not always translate into straightforward predictions of intervention impact when deployed on larger scales or in contexts that differ from that of a trial. This first challenge is one of understanding infectious disease burden and the second is one of predicting how infectious diseases will respond to interventions. Our research is organized around addressing these two challenges.
Mechanistic models are the core of our approach. Whereas statistical models capture phenomenological associations among variables (e.g., Y depends linearly on X), the structure and parameters of mechanistic models are derived from biological principles and assumptions (e.g., the rate at which susceptible individuals become infected depends on the rate at which they come into contact with infectious individuals and the probability that transmission occurs given contact). To address the first challenge of understanding infectious disease burden, we exploit the fact that models can synthesize information about multiple components of a system and generate new understanding of biological quantities that often cannot be observed directly. To address the second challenge of predicting how infectious diseases will respond to interventions, we leverage the fact that mechanistic models are equipped to explain differences in disease burden in space, time, and in other respects, which enables predictions to be made about intervention impacts in contexts that cannot be observed directly. Across the diverse set of projects in our lab, we employ mechanistic models with a variety of forms (ranging from simple equation-based models to complex agent-based models) and complement them with statistical approaches as needed. We frequently collaborate with empiricists to enrich the relevance and impact of our work.
