Reproducibility package for dynamic-t multimodal landmark survival prediction in multiple myeloma: MMRF CoMMpass, external validation, and training outputs

This Figshare item contains the reproducibility package for a dynamic-t multimodal landmark survival prediction framework for multiple myeloma. The codebase supports multimodal survival modeling using laboratory time-series, drug exposure, and imaging-derived features, together with landmark-based evaluation and reporting of C-index, AUC@1y, and AUC@2y.
The deposited materials include processed MMRF CoMMpass resources, external validation resources, and training/evaluation outputs. These files support reproducible model training, benchmarking, visualization, and external validation.