sleev: An R Package for Semiparametric Likelihood Estimation with Errors in Variables

Area of Science:

• Biomedical research
• Statistical methodology
• Data science

Background:

• Routinely collected data in biomedical research often contain measurement errors in outcomes or covariates.
• Two-phase study designs are common, where only a subsample of data is validated.
• Analyzing error-prone data requires specialized statistical methods.

Purpose of the Study:

• To address the need for computationally efficient and user-friendly tools for analyzing error-prone data in two-phase studies.
• To introduce the R package `sleev` for implementing the sieve maximum likelihood estimator (SMLE).
• To facilitate semiparametric likelihood-based inference for error-prone binary and continuous outcomes and covariates.

Main Methods:

• Utilized the sieve maximum likelihood estimator (SMLE) approach.
• Developed the R package `sleev` to implement SMLE for two-phase studies.
• The package handles error-prone binary and continuous outcomes and covariates.

Main Results:

• The R package `sleev` provides a user-friendly tool for applying SMLE.
• Enables efficient and robust analysis of complex error-prone data.
• Supports analysis for both binary and continuous outcomes with measurement error.

Conclusions:

• The `sleev` R package effectively fills the gap for analyzing error-prone data in two-phase studies.
• It enhances the accessibility and efficiency of using SMLE in biomedical research.
• The tool supports a wide range of data types, including error-prone responses and covariates.