Document is current

Article History

Received: 8 September 2025

Accepted: 20 January 2026

First Online: 28 January 2026

Declarations

:

: This study does not involve any human participants, animals, or sensitive personal data. Therefore, ethical approval was not required.

: All authors have reviewed the final version of the manuscript and give their full consent for publication.

: The authors declare no competing interests.

: To facilitate reproducibility, the dataset source, preprocessing steps, model hyperparameters, federated learning configuration, and evaluation protocols are fully described in the manuscript. Experiments were conducted using a fixed patient-level train–test split, and all preprocessing operations were fitted on training data only. Model training was repeated across multiple runs with different random seeds to assess robustness. Software versions and implementation details are explicitly reported to enable independent replication of the experimental workflow.