Benchmark Dataset

Module Introduction

The "Benchmarking Datasets" module serves as a repository for rigorous validation, hosting three distinct reference libraries: the Real IPD Dataset, the Published KM Dataset, and the Synthetic KM Dataset.

The complete benchmarking datasets are publicly accessible via Zenodo at https://zenodo.org/records/18320575.

Each dataset serves a specific validation purpose:

• Real IPD Dataset: Designed to assess distributional fidelity. It validates the concordance of the Weibull distribution parameters between the reconstructed IPD and the ground-truth IPD.
• Published KM Dataset & Synthetic KM Dataset: Focused on statistical precision. These datasets are utilized to evaluate the accuracy of survival statistics derived from the reconstructed IPD by comparing them against the ground-truth values calculated from the original data.