Model-based integration of wastewater-based surveillance into pandemic monitoring

ABOUT THE SPEAKER
Daniele Proverbio is a postdoctoral researcher at the University of Trento, under the ERC INSPIRE Project. He’s currently working on developing methods to assess robustness and resilience of complex systems, with applications to biology and epidemiology. He holds a PhD in Computational Sciences from the University of Luxembourg, an MBA from Collège des Ingénieurs and a degree in Physics of Complex Systems from the University of Turin. He has also served in the Research Luxembourg COVID-19 taskforce.
His broader interests embrace the use and impact of complex systems models in societies, including mathematical models and AI, with a focus on robustness, interpretability and ethical

ABSTRACT
To track epidemic dynamics, wastewater-based epidemiology (WBE) is emerging as a powerful complement to population sampling, identifying increases or decreases of pathogenic genetic material in wastewaters to infer trends in epidemic diffusion. However, integrating wastewater-based epidemiology into pandemic surveillance requires making it quantitative, and combining predictions from various data sources.
To this end, I will discuss a causal model to quantify the infectious shedding population based on wastewater data. The model, termed CoWWAn (COVID-19 Wastewater Analyser), embeds an epidemiological SEIR model, augmented by suitable compartments to represent viral shedding into wastewater, within a Kalman filter, to routinely estimate rate parameters and calibrate the models to reproduce observed data. The model is designed to handle noisy data as inputs and, if both wastewater data and case numbers are available, it allows integrating both data sources to further improve the prediction of future pandemic trends.
After testing on several global data sources, CoWWAn successfully reconstructed case numbers from data and allowed predictions. Its performance allows to use wastewater data to complement case numbers in epidemic modelling, forecasting, scenario modelling and outbreak anticipation. By combining both data sources, the forecasting precision further improves.
The CoWWAn model thus constitutes a first step towards multi-signal modeling of infectious diseases, to improve quantitative monitoring of epidemics.