LimeTr

LimeTr implements robust mixed-effects modeling by integrating trimming into the marginal likelihood to provide robust estimates for analyses with outliers, nonlinear observation models, explicit priors, and constrained dose-response relationships (e.g., monotonicity and concavity) in longitudinal analyses and meta-analyses.

Key Features:

• Robust estimation with trimming: Incorporates trimming within the marginal likelihood of mixed-effects models to reduce the influence of outliers and improve reliability in longitudinal analyses and meta-analyses.
• Nonlinear observation models and priors: Supports nonlinear measurement/observation models and explicit priors beyond linear random-effects approximations such as log-linear models.
• Constraints on dose-response relationships: Allows imposition of constraints including monotonicity and concavity on dose-response functions.
• Computational efficiency: Provides computational performance improvements relative to other robust alternatives for large-scale analyses.

Scientific Applications:

• Meta-analysis: Produces more accurate pooled estimates in the presence of outliers.
• Longitudinal studies: Handles complex repeated-measures structures and outlier robustness in longitudinal analyses.
• Global health research: Enables constrained modeling of dose-response relationships relevant to health and epidemiological analyses.

Methodology:

Integrates trimming into the marginal likelihood of mixed-effects models and supports nonlinear observation models, explicit priors, and constraints such as monotonicity and concavity on dose-response relationships.