Reproducibility materials for cross-national forecasting of disability and death from longitudinal ageing trajectories
Description
This dataset contains the public reproducibility materials for a cross-national study forecasting transitions between independent living, instrumental activities of daily living limitation, activities of daily living limitation, and death using longitudinal ageing trajectories. The study used harmonised data from six international ageing cohorts: the Health and Retirement Study (HRS), English Longitudinal Study of Ageing (ELSA), Survey of Health, Ageing and Retirement in Europe (SHARE), China Health and Retirement Longitudinal Study (CHARLS), Korean Longitudinal Study of Ageing (KLoSA), and Mexican Health and Aging Study (MHAS). HRS, ELSA, and SHARE were used for model development, whereas CHARLS, KLoSA, and MHAS were retained as fully held-out external validation cohorts. The archive includes analysis and modelling code, machine-readable configuration files, computational environment specifications, synthetic examples, aggregate audit tables, summary results, figures, checksum manifests, and detailed instructions for independently obtaining the source cohort products and reconstructing the analytical datasets. It supports reproduction of conventional machine-learning benchmarks, longitudinal neural-network comparisons, Qwen3-8B QLoRA experiments, temporal ablations, training-scale analyses, missingness and cohort-fingerprint audits, calibration analyses, decision-curve analyses, subgroup evaluations, and participant-clustered paired bootstrap comparisons. Participant-level source data, derived microdata, individual-level predictions, fitted model objects, and model weights are not included because the original cohort data-use agreements do not permit redistribution by the authors. Researchers must obtain the relevant source data directly from the respective cohort custodians and comply with their registration and data-use requirements. The included reconstruction documentation specifies the required data products, variable harmonisation procedures, participant-level partitioning rules, fixed random seeds, model configurations, and integrity checks. These materials are intended to support transparent methodological evaluation and independent reproduction of the reported analyses. They are not intended for clinical use or individual-level risk assessment.