SMART 2.0 — clinical assessment instrument for tracheal pathology: validation supplement and analysis pipeline (v1.1.2)

This deposit contains the analysis pipeline and de-identified supplementary materials for the validation study of SMART 2.0 — a clinical assessment instrument for tracheal pathology (cicatricial tracheal stenosis). The scale comprises five domains (S — Stenosis history, M — Multiple tracheostomies, A — Anatomical complexity, R — Recurrent course, T — Tissue quality), total range 0–20 points, weights expert-derived. Package contents: supplementary tables S1–S15 (cohort flow, characteristics, primary and secondary hypothesis tests, pair-level transitions, whole-track trajectory, subgroup analysis, domain breakdown, era stratification, alternative outcome metrics, restenosis quality decomposition); OSF pre-registration v1.1 (English) and mini-amendments v1.1.1 (environment correction) and v1.1.2 (refined classification); root and inventory documents; Python and Node.js scripts (numbered 06–36 by order of execution); environment lock file. Key results: analytic cohort n = 1209; primary hypothesis H1 (construct validity) supported by Jonckheere–Terpstra one-sided trend test, z = 9.26, p < 0.001, with monotonic medians 6, 7, 8, 9 across four treatment strategy categories; secondary predictive analyses for composite and end-state outcomes yielded AUC 0.60 and 0.69 respectively (DeLong 95% CI). Raw patient-level data are not included due to regulatory constraints on personal health information; access may be requested via the corresponding author with documented IRB approval. All implementations of statistical tests (Jonckheere–Terpstra, Cochran–Armitage trend, Dunn with Holm correction, DeLong AUC CI) are standalone functions based on scipy.stats and sklearn.metrics. See README.md inside the archive for the full description.