Scholarlink

<p>Public transport drivers are pivotal to the functioning of modern transportation systems. Recently, the mental health of these drivers, particularly in China where the workforce is extensive, has garnered significant attention due to the profound societal impact of their well-being.&#xa0;This study aims to employ machine learning techniques to predict depression risk among public transport drivers and to investigate the determinants of their depressive states.&#xa0;We analyzed demographic, personality, and psychological data from 2,442 drivers in Jiangsu Province, China, using five machine learning algorithms: Random Forest, Gradient Boosting Machine, Support Vector Machine, Logistic Regression with Lasso Regularization, and standard Logistic Regression. The study also evaluated the influence of four feature selection methods on the performance of these models.&#xa0;The Gradient Boosting Machine outperformed other models in terms of overall accuracy. Recursive Feature Elimination was the most effective feature selection method, substantially enhancing model performance. Key predictors of depression included phobic anxiety, neuroticism, perceived stress, general anxiety, and hostility.&#xa0;Machine learning approaches, notably the Gradient Boosting Machine, demonstrate high precision in predicting depression risks among public transport drivers.</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

Predicting depression risk in Chinese public transit drivers using machine learning algorithms

  • 期刊论文

  • 发表日期：

    2025年4月28日

查看全文
  • Shuliang Bai,
  • Peibing Liu,
  • Bing Zhang,
  • Renlai Zhou

摘要

Public transport drivers are pivotal to the functioning of modern transportation systems. Recently, the mental health of these drivers, particularly in China where the workforce is extensive, has garnered significant attention due to the profound societal impact of their well-being. This study aims to employ machine learning techniques to predict depression risk among public transport drivers and to investigate the determinants of their depressive states. We analyzed demographic, personality, and psychological data from 2,442 drivers in Jiangsu Province, China, using five machine learning algorithms: Random Forest, Gradient Boosting Machine, Support Vector Machine, Logistic Regression with Lasso Regularization, and standard Logistic Regression. The study also evaluated the influence of four feature selection methods on the performance of these models. The Gradient Boosting Machine outperformed other models in terms of overall accuracy. Recursive Feature Elimination was the most effective feature selection method, substantially enhancing model performance. Key predictors of depression included phobic anxiety, neuroticism, perceived stress, general anxiety, and hostility. Machine learning approaches, notably the Gradient Boosting Machine, demonstrate high precision in predicting depression risks among public transport drivers.