MetaHD - A Multivariate Meta-Analysis Model for High-Dimensional Data

Performs multivariate meta-analysis for high-dimensional data to integrate and collectively analyse individual-level data from multiple studies, as well as to combine summary estimates. This approach accounts for correlation between outcomes, incorporates within- and between-study variability, handles missing values, and uses shrinkage estimation to accommodate high dimensionality. The 'MetaHD' R package provides access to our multivariate meta-analysis approach, along with a comprehensive suite of existing meta-analysis methods, including fixed-effects and random-effects models, Fisher's method, Stouffer's method, the weighted Z method, Lancaster's method, the weighted Fisher's method, and vote-counting approach. Visualisation tools are provided for interpreting and comparing results across methods, including Venn diagrams, UpSet plots, and ROC curves, heatmaps of pooled effect sizes and correlations among outcomes. A detailed vignette with example datasets and code for data preparation and analysis is available at <https://alyshadelivera.github.io/MetaHD_vignette/>.