A functional inference for multivariate current status data with mismeasured covariate

Area of Science:

• Biostatistics
• Epidemiology
• Health Data Science

Background:

• Current status failure time data analysis has addressed covariate measurement error, but multivariate extensions remain unexplored.
• Health surveys often involve multiple correlated health outcomes (e.g., hyperglycemia, hypertension, hyperlipidemia) subject to current status censoring.
• Covariates like body mass index in health surveys can be prone to self-reporting and measurement error.

Purpose of the Study:

• To propose a functional inference method for multivariate current status data with mismeasured covariates.
• To address correlated failure times and covariate measurement error simultaneously in health survey data.
• To extend existing methods for current status data to a multivariate setting with measurement error.

Main Methods:

• Utilized a proportional odds model for multivariate current status data.
• Employed the working independence strategy to handle correlated observations within subjects.
• Applied the conditional score approach to manage mismeasured covariates without distributional assumptions.
• Combined Newton-Raphson and self-consistency algorithms for stable computation.

Main Results:

• Developed a novel statistical framework for analyzing complex health data.
• Established asymptotic theory for the proposed estimation method.
• Demonstrated the method's stability and applicability through simulation studies and real-world data analysis (three-hypers dataset).

Conclusions:

• The proposed functional inference method effectively handles multivariate current status data with mismeasured covariates.
• The method provides a robust approach for analyzing correlated health outcomes in the presence of covariate error.
• This work offers a valuable tool for researchers analyzing complex health survey data.