SPINEREG reproducibility guide

Supplementary File S2 for Time Course and Individual Heterogeneity of Disability Recovery After Spine Surgery: A Secondary Analysis of the SPINEREG Cohort.

The ZIP contains an executed Jupyter notebook, frozen Python analysis and figure scripts, analysis-ready datasets, a data dictionary, the frozen statistical analysis plan, QC and deviation records, locked machine-readable results, dependency specifications, publication figures, a verification record and SHA256SUMS.txt.

Prerequisites

Python 3.13.5, as recorded by the package verification.
The exact packages in reproducibility/requirements.txt, or the Conda environment in reproducibility/environment.yml.
Jupyter for the notebook workflow.

Reproduce the analysis

conda env create -f reproducibility/environment.yml
conda activate spinereg-reproducibility
jupyter lab reproducibility/SPINEREG_secondary_analysis.ipynb

Run all notebook cells from top to bottom. The notebook writes results to reproducibility/SPINEREG_notebook_results/, checks them against results_tables/, and invokes the publication-figure script. The frozen provenance script can also be run from the reproducibility directory with python SPINEREG_secondary_analysis.py.

Expected result

The packaged record reports: PASS: all frozen numerical outputs reproduced successfully. Expected headline values are N = 1,392; 254 complete postoperative cases; primary mixed-model convergence; Hb × time p = 0.9513112016937315; diagnosis × time p = 0.44235422319823714.

Integrity

The JMNS pre-release audit on 16 October 2026 found zero SHA-256 mismatches. A fresh rerun with the pinned scientific libraries under Python 3.12 reproduced all 13 CSV tables; three floating-point fields differed from the Python 3.13.5 locked outputs by at most 1.47 × 10⁻⁷. Use the recorded Python 3.13.5 environment for byte/tolerance parity.

Provenance

Source workbook SHA-256: 03c0f51af84bc4e8c2fbcf8deef52993d6abb14057a236928a9e47eb1daa4e3f.

CODEDownload Reproducibility Package S2 (ZIP)