Initial CRAN release
- Provides likelihood-based meta-analysis through generalized linear
mixed-effects models using aggregate data.
- Supports covariate adjustment and non-normal response families, with
computationally efficient likelihood evaluation.
- Includes likelihood-based confidence intervals and the fast and
quasi-Monte Carlo integration methods described by Hanada and Sugimoto
(2026).
- Draws selected Wald, profile, SBC, and plug-in prediction intervals
as forest-plot diamonds, adds numeric annotations, and automatically
reserves space for long labels and headers.
- Lets
forest(fit) automatically use safe study-level
data and a uniquely identified treatment coefficient, with SBC as the
default when available.
- Expands the binary, Poisson, and Gaussian vignettes through
forest-plot workflows and replaces the custom Gaussian vignette example
with an overdispersed negative-binomial rate model.
- Supports profile and simple Bartlett-corrected confidence intervals
for custom families when heterogeneity is estimated and an appropriate
study-level sampling variance is supplied through
vi.