Scholarlink

Background <p>The mental health morbidities of the health care workers were long-standing and further intensified due to COVID-19. Although the majority of research focuses on burnout and distress, little emphasis has been placed on the proactive prediction of psychological resilience.</p> Methods <p>Using the nationally-representative “How Right Now Mental Health &amp; Coping” dataset (n = 2055, 2021–2022), we trained machine learning models—Logistic Regression, Random Forest, and Support Vector Machine—for resilience level classification based on psychometric indices, emotional symptoms, sociodemographic characteristics, and coping behaviors. The model performance was assessed with ROC-AUC, precision, recall, and F1-score with cross-validation.</p> Results <p>Logistic Regression demonstrated superior discriminative performance (ROC AUC = 0.816, accuracy = 75.6%) compared to Random Forest and SVM. Stress, depression, and anxiety were the most important predictors of low resilience according to feature importance analysis. Coping mechanisms as protective factors: A linear association between the number of coping strategies (social support, hobbies, prayer, meditation) and resilience scores was found, suggesting a cumulative protective effect.</p> Conclusions <p>Our work presents an interpretable AI model to predict psychosocial resilience in healthcare workers. These results highlight the importance of deploying easy-to-use collection tools to monitor mental distress over time and reveal actionable factors that are predictive of it, guiding personalized mental health interventions in the early stages for sustainable staffing in highly stressful clinical settings.</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 psychosocial resilience in healthcare workers during COVID-19 using interpretable machine learning

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月2日

查看全文
  • Pablo Torres-Carrion,
  • Alicia Zavala Calahorrano,
  • Hector F Gomez A.,
  • Alex Criollo Rodriguez,
  • Olivia Altamirano Guerrero,
  • Susana A Arias T

摘要

Background

The mental health morbidities of the health care workers were long-standing and further intensified due to COVID-19. Although the majority of research focuses on burnout and distress, little emphasis has been placed on the proactive prediction of psychological resilience.

Methods

Using the nationally-representative “How Right Now Mental Health & Coping” dataset (n = 2055, 2021–2022), we trained machine learning models—Logistic Regression, Random Forest, and Support Vector Machine—for resilience level classification based on psychometric indices, emotional symptoms, sociodemographic characteristics, and coping behaviors. The model performance was assessed with ROC-AUC, precision, recall, and F1-score with cross-validation.

Results

Logistic Regression demonstrated superior discriminative performance (ROC AUC = 0.816, accuracy = 75.6%) compared to Random Forest and SVM. Stress, depression, and anxiety were the most important predictors of low resilience according to feature importance analysis. Coping mechanisms as protective factors: A linear association between the number of coping strategies (social support, hobbies, prayer, meditation) and resilience scores was found, suggesting a cumulative protective effect.

Conclusions

Our work presents an interpretable AI model to predict psychosocial resilience in healthcare workers. These results highlight the importance of deploying easy-to-use collection tools to monitor mental distress over time and reveal actionable factors that are predictive of it, guiding personalized mental health interventions in the early stages for sustainable staffing in highly stressful clinical settings.