Scholarlink

<p>The study identifies the patterns in gender specific population units, examines the underlying latent factors, for analyzing the consequent mental health challenges. Such gender demarcated analysis enables policymakers to initiate targeted interventions; therein lies the uniqueness of this study. In this context, a survey of 418 individuals’ data (mostly from the student community) from inside and outside India regarding their mental health status is utilized. This study investigates the application of <i>Latent Class Regression Analysis</i> and <i>Bayesian Structural Equation Modeling (BSEM)</i> over latent class analysis and traditional structural equation modeling, to address the problem. Using multiple clustering indices and fit statistics, a two-class solution best described both male and female populations. In addition, Kernel SHAP values with bootstrap stability analysis are integrated to achieve the feature importance for providing a detailed understanding of the latent factors influencing stress levels across diverse populations. SHAP-informed priors are incorporated into BSEM, which improves the model fit as compared to weak priors (lower WAIC, higher PPPI). Results reveal distinct stress profiles linked to social media use, loneliness, sleep quality, and financial stress, providing a basis for targeted psychological and behavioral interventions. Findings emphasize internally driven stress dynamics and support SHAP-guided prior specification for targeted, subgroup-specific intervention strategies.</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

A gender-specific study for exploring the factors affecting mental health through identifying the underlying patterns and sub-patterns

  • 期刊论文

  • 发表日期：

    2025年10月8日

查看全文
  • Sushovon Jana,
  • Anwesha Sengupta,
  • Prasanta Narayan Dutta,
  • Indranil Mukherjee

摘要

The study identifies the patterns in gender specific population units, examines the underlying latent factors, for analyzing the consequent mental health challenges. Such gender demarcated analysis enables policymakers to initiate targeted interventions; therein lies the uniqueness of this study. In this context, a survey of 418 individuals’ data (mostly from the student community) from inside and outside India regarding their mental health status is utilized. This study investigates the application of Latent Class Regression Analysis and Bayesian Structural Equation Modeling (BSEM) over latent class analysis and traditional structural equation modeling, to address the problem. Using multiple clustering indices and fit statistics, a two-class solution best described both male and female populations. In addition, Kernel SHAP values with bootstrap stability analysis are integrated to achieve the feature importance for providing a detailed understanding of the latent factors influencing stress levels across diverse populations. SHAP-informed priors are incorporated into BSEM, which improves the model fit as compared to weak priors (lower WAIC, higher PPPI). Results reveal distinct stress profiles linked to social media use, loneliness, sleep quality, and financial stress, providing a basis for targeted psychological and behavioral interventions. Findings emphasize internally driven stress dynamics and support SHAP-guided prior specification for targeted, subgroup-specific intervention strategies.