Plan selection sensitivity data and code

Aggregate data and Python workflows for evaluating how the choice of one candidate radiotherapy plan per patient affects paired CT radiomics–dosiomics analyses. The package includes feature- and association-level numerical outputs, a data dictionary, reproducibility scripts, a source-archive checksum manifest, validation reports, and public digital-phantom controls. The included-data workflow uses Python 3.10 or later without additional packages to regenerate numerical summaries, policy contrasts and plots, with integrity checks. A portable CSV workflow supports features obtained with other extraction tools. The full source-analysis workflow specifies preprocessing, paired feature extraction, repeated plan selection and patient-bootstrap analyses using pinned dependencies. Version 1.1.0 adds aggregate results and code for candidate-set structure, prescription and target coverage, moving versus fixed categorisation thresholds, dependence on the number of draws, technique-coherent policy contrasts, simultaneous intervals, influence analyses, and a target-coverage subset. It also includes scripts to regenerate the expanded figures and check the new numerical summaries. The original analysis files are retained. Original DICOM images and patient-level feature matrices are not included. Recalculation of the empirical Monte Carlo and bootstrap analyses requires the four exact external source archives identified in the package and remains subject to their access conditions. The supplied aggregate data support regeneration of summaries and visualisations, rather than patient-level reanalysis. Third-party control-data licences and attribution notices are retained. Version 1.1.0, released 31 October 2026. See README.md and DATA_DICTIONARY.md in the archive for instructions and field definitions.