Comparing and combining data across multiple sources via integration of paired-sample data to correct for measurement...

Area of Science:

• Biostatistics
• Biomedical Research
• Data Science

Background:

• Biomedical studies, including vaccine development, often pool data from multiple laboratories.
• Variations in measurement error between laboratories can compromise data integrity when combining samples.
• Adjusting for inter-laboratory measurement variability is crucial for accurate analysis.

Purpose of the Study:

• To develop and validate methods for adjusting and combining data from multiple laboratories with varying measurement errors.
• To enable robust comparisons of independent samples across different data sources.
• To provide accurate sample size calculations for studies involving inter-laboratory comparisons.

Main Methods:

• Normalization of individual-level data to a common scale using conditional expectations.
• Comparison of mean assay values between independent samples, accounting for inter-source measurement error.
• Development of sample size calculation methods for paired-sample studies to control hypothesis testing error rates.

Main Results:

• Proposed methods effectively normalize data and account for inter-source measurement error.
• Simulation studies demonstrate satisfactory performance across various scenarios with differing measurement errors.
• The methods do not require knowledge of true underlying measurements in external data.

Conclusions:

• The developed statistical methods provide a robust framework for combining and analyzing data from multiple laboratories in biomedical research.
• Accurate adjustment for measurement error enhances the reliability of study outcomes, particularly in vaccine and cancer research.
• The approach is applicable to real-world data, as demonstrated with enzyme-linked immunosorbent spot assay data from HIV vaccine laboratories.