Novel machine learning models including adipose-muscle indices for the prediction of musculoskeletal complications in rheumatoid arthritis: A cross-sectional study
Sage Open Chronic Disease · 2026-02 · 17
DOI: 10.1177/27558428261484147 ↗Kurz gefasst
Querschnittstudie in der Zeitschrift Sage Open Chronic Disease, online erschienen am 28.08.2026. Verfasst von Jing-Tian Shi und 6 weiteren Autor:innen. Beteiligte Einrichtung: Department of Rheumatology and Immunology, The First Affiliated Hospital of Anhui Medical University, Hefei, Anhui, China. Die Arbeit nennt 35 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0). Gefördert von: Major Research Plan.
Abstract (Originalsprache)
Background
This research aimed to investigate the associations of multiple adipose-muscle indices with secondary osteoporosis (OP) and sarcopenia in rheumatoid arthritis (RA) patients.
Objectives
To develop a comprehensive early diagnostic model for musculoskeletal loss in RA patients based on non-invasive adipose-muscle indices.
Design
A cross-sectional and Mendelian randomization (MR) study.
Methods
Fourteen machine learning (ML) algorithms were applied to select features from 33 candidate variables and were incorporated to construct clinical predictive models for OP, sarcopenia, and osteo-sarcopenia (OS). Restricted cubic spline (RCS) analysis explored dose-response relationships. MR methods were used to analyze potential causal effects.
Results
Six key features were identified, of which the fat-to-muscle ratio (FMR) and the skeletal muscle mass-to-visceral fat area ratio (SVR) were the most prominent. FMR was a risk factor for OP, whereas SVR was a protective factor against sarcopenia. The LightGBM model demonstrated superior performance in predicting OS (AUC = 0.988). MR analysis further indicated that body fat mass fully mediated the causal pathway from RA to OP and sarcopenia.
Conclusion
ML models based on hybrid adipose-muscle indices demonstrated exceptional predictive value for subclinical musculoskeletal loss in RA patients.
Schlagwörter
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Department of Rheumatology and Immunology, The First Affiliated Hospital of Anhui Medical University, Hefei, Anhui, China.
- Department of Hematology and Rheumatology, Deyu Medical Ma'anshan General Hospital, Ma'anshan, Anhui, China.
ORCID-Profile: Sheng-Qian Xu
Förderung
- Major Research Plan
Zitieren
APA
Shi, J. T., Wu, T. T., Mo, X. L., Xia, X. X., Chu, Y. R., Wang, J. X., & Xu, S. Q. (2026). Novel machine learning models including adipose-muscle indices for the prediction of musculoskeletal complications in rheumatoid arthritis: A cross-sectional study. Sage Open Chronic Disease, 17. https://doi.org/10.1177/27558428261484147
Vancouver
Shi JT, Wu TT, Mo XL, Xia XX, Chu YR, Wang JX, et al. Novel machine learning models including adipose-muscle indices for the prediction of musculoskeletal complications in rheumatoid arthritis: A cross-sectional study. Sage Open Chronic Disease. 2026;17. doi:10.1177/27558428261484147
BibTeX
@article{Shi2026,
author = {Shi, Jing-Tian and Wu, Ting-Ting and Mo, Xiao-Ling and Xia, Xiao-Xuan and Chu, Yi-Ran and Wang, Jian-Xiong and Xu, Sheng-Qian},
title = {Novel machine learning models including adipose-muscle indices for the prediction of musculoskeletal complications in rheumatoid arthritis: A cross-sectional study},
journal = {Sage Open Chronic Disease},
year = {2026},
volume = {17},
doi = {10.1177/27558428261484147}
}
EUVIMED