Regression analysis for multiple-disease group testing data

Area of Science:

• Biostatistics
• Epidemiology
• Public Health

Background:

• Group testing is a cost-effective strategy for screening large populations by pooling individual specimens.
• Analyzing group testing data presents challenges due to the need for individual-level inferences from group-level results.
• Simultaneous testing for multiple diseases introduces complexity due to potential correlations in individual disease statuses.

Purpose of the Study:

• To develop advanced regression techniques for analyzing multiple-disease group testing data.
• To provide a method for accurate individual disease status inference when only group test results are available.
• To address the challenge of correlated individual disease statuses in simultaneous group testing scenarios.

Main Methods:

• Proposed novel regression techniques specifically designed for multiple-disease group testing.
• Developed an expectation-solution based algorithm for parameter estimation and inference.
• The algorithm ensures consistent parameter estimates and facilitates natural large-sample inference procedures.

Main Results:

• The proposed methodology yields consistent parameter estimates for individual disease statuses.
• The developed algorithm supports robust large-sample inference procedures.
• Successful application demonstrated on real-world datasets for chlamydia, gonorrhea, and prenatal infectious diseases.

Conclusions:

• The new regression techniques effectively handle the complexities of multiple-disease group testing.
• The methodology offers a significant advancement for cost-effective disease screening and surveillance.
• This approach enhances the ability to infer individual health statuses from pooled testing data.