Scholarlink

Background <p>Sleep disturbances and depressive symptoms frequently co-occur in older adults. Both conditions follow distinct, time-varying trajectories. Nevertheless, most studies rely on cross-sectional assessments, limiting evidence regarding their joint longitudinal evolution and associations with incident chronic diseases.</p> Methods <p>This study utilized data drawn from 3,221 participants (aged ≥ 60 years) enrolled in the China Health and Retirement Longitudinal Study (CHARLS). Group-based multi-trajectory models (GBMTM) were constructed using repeated measures from 2011 to 2018 to identify heterogeneous joint trajectories of sleep duration and depressive symptoms. Cox proportional hazards models assessed associations with 13 incident chronic diseases and multimorbidity. Additionally, a machine learning framework incorporating seven algorithms was applied to identify baseline predictors of high-risk trajectories, followed by SHAP analysis to enhance model interpretability.</p> Results <p>The mean age of participants was 65.80 ± 4.93 years. We identified four joint trajectories: normal-stable sleep and low-stable depression (24.46%), short-stable sleep and low-stable depression (27.17%), normal-increasing sleep and moderate-increasing depression (25.00%), and short-decreasing sleep and high-increasing depression (23.38%). The “short-decreasing sleep and high-increasing depression” trajectory exhibited the highest risks, notably for memory-related disorders (HR = 3.08), stroke (HR = 2.56), and multimorbidity (HR = 1.97). XGBoost and ANN achieved the best predictive performance (AUC = 0.805), with body pain and cognitive function identified as primary predictors.</p> Conclusion <p>The trajectory characterized by declining sleep duration and worsening depressive symptoms was associated with heightened risks of multimorbidity and various chronic conditions in older adults. These findings underscore the necessity of integrating sleep and depressive symptom surveillance for chronic disease prevention. Furthermore, early screening for body pain and cognitive decline may facilitate the timely identification of high-risk individuals and inform targeted precision interventions.</p> Graphical Abstract <p></p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Joint trajectories of sleep duration and depressive symptoms and risk of incident multimorbidity: a longitudinal analysis with machine learning prediction

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月28日

查看全文
  • Jiecheng Jiang,
  • Zhujiang Li,
  • Zhuo Zhang,
  • Shiyu Ji,
  • Zefeng Zhang,
  • Yixuan Wu,
  • Yaqi Li,
  • Mingyu Yu,
  • Peipei Qiao,
  • Junxiang Xu,
  • Jun Wang,
  • Panpan Huang

摘要

Background

Sleep disturbances and depressive symptoms frequently co-occur in older adults. Both conditions follow distinct, time-varying trajectories. Nevertheless, most studies rely on cross-sectional assessments, limiting evidence regarding their joint longitudinal evolution and associations with incident chronic diseases.

Methods

This study utilized data drawn from 3,221 participants (aged ≥ 60 years) enrolled in the China Health and Retirement Longitudinal Study (CHARLS). Group-based multi-trajectory models (GBMTM) were constructed using repeated measures from 2011 to 2018 to identify heterogeneous joint trajectories of sleep duration and depressive symptoms. Cox proportional hazards models assessed associations with 13 incident chronic diseases and multimorbidity. Additionally, a machine learning framework incorporating seven algorithms was applied to identify baseline predictors of high-risk trajectories, followed by SHAP analysis to enhance model interpretability.

Results

The mean age of participants was 65.80 ± 4.93 years. We identified four joint trajectories: normal-stable sleep and low-stable depression (24.46%), short-stable sleep and low-stable depression (27.17%), normal-increasing sleep and moderate-increasing depression (25.00%), and short-decreasing sleep and high-increasing depression (23.38%). The “short-decreasing sleep and high-increasing depression” trajectory exhibited the highest risks, notably for memory-related disorders (HR = 3.08), stroke (HR = 2.56), and multimorbidity (HR = 1.97). XGBoost and ANN achieved the best predictive performance (AUC = 0.805), with body pain and cognitive function identified as primary predictors.

Conclusion

The trajectory characterized by declining sleep duration and worsening depressive symptoms was associated with heightened risks of multimorbidity and various chronic conditions in older adults. These findings underscore the necessity of integrating sleep and depressive symptom surveillance for chronic disease prevention. Furthermore, early screening for body pain and cognitive decline may facilitate the timely identification of high-risk individuals and inform targeted precision interventions.

Graphical Abstract