Analysis plan v4 and analysis code (locked before data access): replication in the Korea National Health and Nutrition Examination Survey of the training-label findings from NHANES

Prespecified statistical analysis plan (version 4) and the analysis code that implements it, locked on 3 November 2026 and archived before any microdata file of the Korea National Health and Nutrition Examination Survey (KNHANES) was downloaded for this project; the author had never downloaded KNHANES microdata before. The plan specifies (A) the re-development of 12 classifiers of diabetes and hypertension status (spline logistic regression, explainable boosting machine and CatBoost, each trained on a reported diagnosis or a single measurement) on KNHANES 2013–2018 with the procedure frozen in the NHANES study, and confirmatory tests in KNHANES 2022–2024 of five hypotheses (K1–K5) derived from the NHANES results; and (B) the external validation of the 12 frozen NHANES models in KNHANES 2022–2024. The code was tested only on synthetic data before archiving. SHA-256 of ANALYSIS_PLAN_v4.pdf: eeaa2b92cf98c820874136662c51ec6d563a2a2240bc330b3c50a0cffea5556d. SHA-256 of k4_code_archive.zip: b2cf05d263ff9add3ce2c46b7ca7e486b0871d999a9c97124742bce813f7c2db. Related records: analysis plan v3 for NHANES 2021–2023 (https://doi.org/10.5281/zenodo.23417505) and the earlier project archive (https://doi.org/10.5281/zenodo.23303323).