Scholarlink

Background <p>Chronic Obstructive Pulmonary Disease (COPD) is a prevalent respiratory condition often accompanied by depression, which exacerbates disease burden and impairs quality of life. Early identification of depression risk in COPD patients remains a clinical challenge.</p> Objective <p>This study aimed to develop a machine learning-based model to predict depression risk in COPD patients, utilizing interpretable features from clinical and demographic data to support early intervention.</p> Methods <p>Data were extracted from the National Health and Nutrition Examination Survey (NHANES), involving 1,638 COPD patients. Depression was assessed using the Patient Health Questionnaire-9 (PHQ-9) scale. Feature selection was performed with Boruta and least absolute shrinkage and selection operator (LASSO) algorithms, identifying key predictors from demographic, lifestyle, medical history, and laboratory variables. Nine machine learning models were trained and evaluated, with performance assessed via accuracy, area under the curve (AUC), calibration, and clinical utility metrics.</p> Results <p>Significant predictors of depression included sleep disturbances, age, poverty, hypertension, and comorbidities like cardiovascular disease. The Support Vector Machine (SVM) model achieved the highest performance, with an AUC of 0.890 in the validation set and 0.887 in the test set, demonstrating robust discriminative ability and clinical applicability. SHapley Additive exPlanations (SHAP) analysis enhanced model interpretability.&#xa0;Notably, sleep disturbances, younger age, and greater socioeconomic deprivation were associated with an elevated risk of depression.</p> Conclusion <p>This study presents a reliable SVM-based model for predicting depression risk in COPD patients, leveraging NHANES data and interpretable features. It offers a valuable tool for early screening and personalized care, with potential to improve mental health outcomes in this population.</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

Development and validation of a risk prediction model for depression in patients with chronic obstructive pulmonary disease

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年7月3日

查看全文
  • Tong Feng,
  • PeiPei Li,
  • Ran Duan,
  • Zhi Jin

摘要

Background

Chronic Obstructive Pulmonary Disease (COPD) is a prevalent respiratory condition often accompanied by depression, which exacerbates disease burden and impairs quality of life. Early identification of depression risk in COPD patients remains a clinical challenge.

Objective

This study aimed to develop a machine learning-based model to predict depression risk in COPD patients, utilizing interpretable features from clinical and demographic data to support early intervention.

Methods

Data were extracted from the National Health and Nutrition Examination Survey (NHANES), involving 1,638 COPD patients. Depression was assessed using the Patient Health Questionnaire-9 (PHQ-9) scale. Feature selection was performed with Boruta and least absolute shrinkage and selection operator (LASSO) algorithms, identifying key predictors from demographic, lifestyle, medical history, and laboratory variables. Nine machine learning models were trained and evaluated, with performance assessed via accuracy, area under the curve (AUC), calibration, and clinical utility metrics.

Results

Significant predictors of depression included sleep disturbances, age, poverty, hypertension, and comorbidities like cardiovascular disease. The Support Vector Machine (SVM) model achieved the highest performance, with an AUC of 0.890 in the validation set and 0.887 in the test set, demonstrating robust discriminative ability and clinical applicability. SHapley Additive exPlanations (SHAP) analysis enhanced model interpretability. Notably, sleep disturbances, younger age, and greater socioeconomic deprivation were associated with an elevated risk of depression.

Conclusion

This study presents a reliable SVM-based model for predicting depression risk in COPD patients, leveraging NHANES data and interpretable features. It offers a valuable tool for early screening and personalized care, with potential to improve mental health outcomes in this population.