Calibration weighted estimation of semiparametric transformation models for two-phase sampling

Area of Science:

• Biostatistics
• Vaccinology
• Epidemiology

Background:

• Two-phase study designs are crucial for cost-effectively analyzing expensive biomarkers in large cohorts, particularly in vaccine immunology.
• Immune response biomarkers are vital for understanding vaccine effectiveness and protection against viral infections.
• Leveraging readily available correlated variables can improve the efficiency of estimating effects of expensive biomarkers.

Purpose of the Study:

• To develop an improved inverse probability weighted estimation approach for semiparametric transformation models within a two-phase study design.
• To enhance the analysis of costly immune response biomarkers in vaccine trials.
• To provide a robust statistical framework for modeling waning immune responses over time.

Main Methods:

• Developed an improved inverse probability weighted estimation approach.
• Utilized weights calibration, drawing from survey statistics principles.
• Applied semiparametric transformation models, encompassing Cox PH and proportional odds models.
• Derived asymptotic theory for the proposed estimator.

Main Results:

• The improved estimation approach demonstrated enhanced efficiency in simulation studies.
• The method effectively models the effects of immune response biomarkers, accounting for their temporal waning.
• Asymptotic theory was developed to support the statistical validity of the estimator.

Conclusions:

• The proposed weights calibration approach offers a statistically sound and efficient method for analyzing two-phase studies with expensive biomarkers.
• This method is particularly valuable in vaccine immunology for understanding immune responses and vaccine efficacy.
• The approach was successfully illustrated using data from HIV-1 vaccine efficacy trials.