Multilevel Unanchored Meta-Regression for Indirect Treatment Comparisons

Bayesian multilevel unanchored meta-regression (ML-UMR) for indirect treatment comparisons using individual patient data (IPD) and aggregate data (AgD). Implements shared prognostic factor assumption (SPFA) and relaxed SPFA models for binary, continuous, count, and time-to-event (survival) outcomes via 'Stan', including parametric and flexible M-spline survival baselines. Also provides a one-arm simulated treatment comparison (STC) via comparator-population G-computation and naive unadjusted benchmarks. ML-UMR is an adaptation of the ML-NMR methodology (Phillippo et al. 2020, ) implemented in the 'multinma' package (GPL-3) to the unanchored two-trial case. The public API uses familiar names for data setup, integration and effect summaries, but ML-NMR and ML-UMR specifications and fitted objects are not interchangeable because their evidence structures and models differ.