Scholarlink

Background <p>Fatty Liver Disease (FLD) has become the most common chronic liver disease worldwide, posing a great threat to public health. The existing methods are primarily based on clinical patient cohorts, supplemented by urban environmental factors, to elucidate biological mechanisms. However, insufficient geographic information and spatial resolution limit the public risk identification. This study integrates real and virtual multi-source data and proposes an explainable geospatial framework for street-level HS risk assessment.</p> Methods <p>Firstly, based on prior medical knowledge, we construct an indicator system comprising three dimensions: social economy, built environment, and psychology (SBP). Then, the multi-model generalization and contribution voting strategy are used for preliminary analysis and optimization to identify key factors. Finally, based on ablation, SHapley Additive exPlanations (SHAP) and spatial analysis, the independent and synergistic effects of different geographical features are revealed. This study takes 90 streets in Hefei, China, as the research area, and combines multi-regional and dynamic prediction (2023–2024) for verification.</p> Results <p>The results show that the SBP framework achieves good performance in both multi-regional (R<sup>2</sup> = 0.765; MAPE = 0.273) and dynamic prediction (R<sup>2</sup> = 0.75; MAPE = 0.328) scenarios. The analysis shows that socio-economic features such as age structure and medical care are the most fundamental factors. Further, the superposition of built environmental features, such as health space accessibility and traffic conditions, related to the difference in spatial distribution of disease and the role of psychology.</p> Conclusions <p>This study focuses on the challenge of unknown the geospatial factors for HS risk in the context of urbanization, aiming to solve three core issues: insufficient spatial resolution, indicator framework design, and interpretability. The empirical results demonstrate the effectiveness of the SBP framework in identifying spatial heterogeneity of FLD, emphasizes its potential to identify regional differences and inform policy design. The method also provides a quantifiable scientific basis for precise FLD prevention and refined urban health management.</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

An explainable geospatial framework for fine-scale Fatty Liver Disease (FLD) risk assessment: roles of social economy, built environment, and psychology

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月21日

查看全文
  • Qiushi Hu,
  • Huiheng Wang,
  • Chunmei Yang,
  • Jiaxin Wang,
  • Tongxin Xia,
  • Xinmin Chu,
  • Xuehan Liu

摘要

Background

Fatty Liver Disease (FLD) has become the most common chronic liver disease worldwide, posing a great threat to public health. The existing methods are primarily based on clinical patient cohorts, supplemented by urban environmental factors, to elucidate biological mechanisms. However, insufficient geographic information and spatial resolution limit the public risk identification. This study integrates real and virtual multi-source data and proposes an explainable geospatial framework for street-level HS risk assessment.

Methods

Firstly, based on prior medical knowledge, we construct an indicator system comprising three dimensions: social economy, built environment, and psychology (SBP). Then, the multi-model generalization and contribution voting strategy are used for preliminary analysis and optimization to identify key factors. Finally, based on ablation, SHapley Additive exPlanations (SHAP) and spatial analysis, the independent and synergistic effects of different geographical features are revealed. This study takes 90 streets in Hefei, China, as the research area, and combines multi-regional and dynamic prediction (2023–2024) for verification.

Results

The results show that the SBP framework achieves good performance in both multi-regional (R2 = 0.765; MAPE = 0.273) and dynamic prediction (R2 = 0.75; MAPE = 0.328) scenarios. The analysis shows that socio-economic features such as age structure and medical care are the most fundamental factors. Further, the superposition of built environmental features, such as health space accessibility and traffic conditions, related to the difference in spatial distribution of disease and the role of psychology.

Conclusions

This study focuses on the challenge of unknown the geospatial factors for HS risk in the context of urbanization, aiming to solve three core issues: insufficient spatial resolution, indicator framework design, and interpretability. The empirical results demonstrate the effectiveness of the SBP framework in identifying spatial heterogeneity of FLD, emphasizes its potential to identify regional differences and inform policy design. The method also provides a quantifiable scientific basis for precise FLD prevention and refined urban health management.