Beyond functional disability: longitudinal joint trajectories of muscle strength and walking speed and their association with new-onset diabetes risk among older adults without ADL impairment
- 1The First Affiliated Hospital of Wenzhou Medical University
Abstract
Background: With population aging, diabetes poses a growing threat to older adults. However, whether longitudinal trajectories of muscle strength and walking speed can identify diabetes risk before ADL impairment develops remains unclear.
Methods: This longitudinal cohort study used data from the HRS, ELSA, SHARE, and CHARLS. Group-based trajectory modeling (GBTM) was applied to identify joint longitudinal trajectories of walking speed and muscle strength. Associations between trajectory groups and new-onset diabetes were assessed using inverse probability of treatment weighting (IPTW)-adjusted Cox proportional hazards models. In addition, 129 candidate machine learning models, comprising nine individual algorithms and 120 Stacking combinations, were evaluated for new-onset diabetes prediction, and SHAP (SHapley Additive exPlanations) analysis was used to interpret model predictions. The robustness of the findings was further evaluated using competing risk models, discrete-time logistic regression models, stratified analyses, and cross-cohort machine learning validation.
Results: During median follow-up periods of 10, 12, 6, and 8 years, 341, 265, 170, and 448 new-onset diabetes events occurred in the HRS, ELSA, SHARE, and CHARLS cohorts, respectively. GBTM identified distinct joint trajectories of muscle strength and walking speed across cohorts. Compared with participants maintaining higher functional levels, those with unfavorable trajectories had significantly increased risks of incident diabetes in fully adjusted IPTW-Cox models (HRS: adjusted HR = 4.17, 95% CI: 3.10–5.61; ELSA: adjusted HR = 3.85, 95% CI: 2.55–5.79; SHARE: adjusted HR = 3.38, 95% CI: 2.25–5.06; CHARLS: adjusted HR = 2.60, 95% CI: 1.87–3.61). Sensitivity analyses consistently supported these associations. Machine learning models further evaluated the contribution of joint trajectories within diabetes prediction frameworks, with SHAP analyses indicating that these trajectory features contributed to model predictions.
Conclusions: Unfavorable joint trajectories of muscle strength and walking speed were associated with higher incident diabetes risk among older adults without baseline ADL impairment, highlighting the potential value of longitudinal functional monitoring for early diabetes risk stratification.
Related
- Disability trajectories and their association with cognition in older adults: a network analysis based on 7⁃year CHARLS longitudinal data
- Prediction of cognitive impairment in Parkinson’s disease using TabPFN: an integrative analysis of clinical data, longitudinal trajectories, and lipidomics
- Socioeconomic position and functional dependency trajectories across six longitudinal ageing cohorts
- Association Between Disability and Longitudinal Brain Volumetric Measures Across the Peri-Pregnancy Period in Women with Multiple Sclerosis: A Pilot Study
- Edentulism and functional disability in older adults: A longitudinal and cross-national study in England, the United States, and India
BibTeX
@article{chen2026beyond,
title = {{Beyond functional disability: longitudinal joint trajectories of muscle strength and walking speed and their association with new-onset diabetes risk among older adults without ADL impairment}},
author = {Xiangyang Chen and Xuan Wu and Sizhe Feng},
journal = {Frontiers in Endocrinology},
year = {2026},
month = oct,
doi = {10.3389/fendo.2026.1935223},
url = {https://www.frontiersin.org/articles/10.3389/fendo.2026.1935223/full},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Beyond functional disability: longitudinal joint trajectories of muscle strength and walking speed and their association with new-onset diabetes risk among older adults without ADL impairment
AU - Xiangyang Chen
AU - Xuan Wu
AU - Sizhe Feng
PY - 2026
DA - 2026/10/28
JO - Frontiers in Endocrinology
DO - 10.3389/fendo.2026.1935223
UR - https://www.frontiersin.org/articles/10.3389/fendo.2026.1935223/full
AB - Background: With population aging, diabetes poses a growing threat to older adults. However, whether longitudinal trajectories of muscle strength and walking speed can identify diabetes risk before ADL impairment develops remains unclear. Methods: This longitudinal cohort study used data from the HRS, ELSA, SHARE, and CHARLS. Group-based trajectory modeling (GBTM) was applied to identify joint longitudinal trajectories of walking speed and muscle strength. Associations between trajectory groups and new-onset diabetes were assessed using inverse probability of treatment weighting (IPTW)-adjusted Cox proportional hazards models. In addition, 129 candidate machine learning models, comprising nine individual algorithms and 120 Stacking combinations, were evaluated for new-onset diabetes prediction, and SHAP (SHapley Additive exPlanations) analysis was used to interpret model predictions. The robustness of the findings was further evaluated using competing risk models, discrete-time logistic regression models, stratified analyses, and cross-cohort machine learning validation. Results: During median follow-up periods of 10, 12, 6, and 8 years, 341, 265, 170, and 448 new-onset diabetes events occurred in the HRS, ELSA, SHARE, and CHARLS cohorts, respectively. GBTM identified distinct joint trajectories of muscle strength and walking speed across cohorts. Compared with participants maintaining higher functional levels, those with unfavorable trajectories had significantly increased risks of incident diabetes in fully adjusted IPTW-Cox models (HRS: adjusted HR = 4.17, 95% CI: 3.10–5.61; ELSA: adjusted HR = 3.85, 95% CI: 2.55–5.79; SHARE: adjusted HR = 3.38, 95% CI: 2.25–5.06; CHARLS: adjusted HR = 2.60, 95% CI: 1.87–3.61). Sensitivity analyses consistently supported these associations. Machine learning models further evaluated the contribution of joint trajectories within diabetes prediction frameworks, with SHAP analyses indicating that these trajectory features contributed to model predictions. Conclusions: Unfavorable joint trajectories of muscle strength and walking speed were associated with higher incident diabetes risk among older adults without baseline ADL impairment, highlighting the potential value of longitudinal functional monitoring for early diabetes risk stratification.
SN - 1664-2392
ER -