Estimating viral prevalence with data fusion for adaptive two-phase pooled sampling

Area of Science:

• Epidemiology
• Biostatistics
• Infectious Disease Ecology

Background:

• The COVID-19 pandemic underscored the need for efficient pathogen prevalence monitoring in human and animal populations.
• Pooled testing is a valuable tool for prevalence estimation, but traditional methods often require extensive retesting.
• Optimizing second-phase sample allocation is crucial when the primary goal is population prevalence estimation, especially in reservoir hosts.

Purpose of the Study:

• To present a novel data fusion approach for two-phased pooled testing to enhance pathogen prevalence estimation.
• To develop a Bayesian data fusion procedure that integrates pooled and individual samples for robust prevalence inference.
• To provide guidance on implementing efficient first- and second-phase sampling plans using data fusion.

Main Methods:

• A two-phased testing strategy was employed, beginning with pooled samples to estimate initial prevalence.
• A Bayesian data fusion procedure was developed to combine data from pooled and individual samples.
• The approach was designed to optimize the allocation of second-phase samples for efficient prevalence estimation.

Main Results:

• Data fusion procedures demonstrated more efficient prevalence estimation compared to traditional methods using only individual samples or single-phase pooled sampling.
• The proposed method allows for more accurate prevalence estimates with a reduced number of samples.
• The study provides a framework for optimizing sampling plans in epidemiological surveillance.

Conclusions:

• The presented data fusion approach offers a more efficient and sample-sparing method for estimating pathogen prevalence.
• These methods are applicable for monitoring diseases in reservoir hosts and assessing spillover risk to humans.
• The findings support improved strategies for tracking pathogens like SARS-CoV-2 in diverse populations.