Unsupervised Bayesian structural time series for estimation of disease variants from wastewater genome samples
Learning objectives
- Wastewater sequencing data present unique challenges/opportunities for data processing and modelling.
- Variants of Concern (VoCs) for SARS-CoV-2 can be useful for public health decision making, but the phylogenetic definition of a VoC is fundamentally different from what we can use in wastewater.
- Estimation of abundances of VoCs can be found with a linear-modelling-based framework, with covariates based on imperfect definitions of the VoCs.
- Repeated sampling over time/locations allows for the estimation of the definitions of the VoCs with fun statistical modelling techniques.
Speaker biography
Dr. Becker is an Assistant Professor in Data Science at Wilfrid Laurier University, with research focusing on finding latent patterns in data. These patterns can be clusters of mutations in wastewater sequencing samples, spatial dependence in wildland fires, and shot locations in sports analytics.