Positively Correlated Samples Save Pooled Testing Costs

Area of Science:

• Epidemiology
• Biostatistics
• Public Health

Background:

• Group testing offers cost savings for mass testing, particularly for infectious diseases like COVID-19.
• Existing group testing models often assume sample independence, which is unrealistic for contagious diseases with familial transmission.
• Positive correlations in test results within groups can arise due to disease transmission patterns.

Purpose of the Study:

• To demonstrate that positive correlations can enhance cost reduction in group testing for COVID-19.
• To rigorously prove the efficacy of the Dorfman two-stage method under positive correlation.
• To develop a novel algorithm for pooled testing that leverages social network structures.

Main Methods:

• Mathematical modeling to analyze the impact of positive correlation on group testing.
• Application and rigorous proof of the Dorfman two-stage method for correlated samples.
• Development of a hierarchical agglomerative algorithm using social graphs for pooled testing.

Main Results:

• Positive correlation between samples within a group can lead to further cost reductions using the Dorfman two-stage method.
• The proposed hierarchical agglomerative algorithm, utilizing social graphs, achieved 20-35% cost reduction compared to random pooling.
• The algorithm effectively incorporates social contact information into the pooling strategy.

Conclusions:

• Accounting for positive correlations in group testing is crucial for optimizing COVID-19 mass testing strategies.
• The Dorfman two-stage method is robust and can be further optimized with correlated data.
• Social graph-informed pooled testing offers a significant advancement in cost-effective disease surveillance.