Scholarlink

<p>Microplastic contamination in urban water systems poses emerging risks to human health, yet integrated frameworks linking sources, exposure pathways, and health outcomes remain limited. This study develops an integrated machine learning–Bayesian–structural equation modeling framework to quantify and predict microplastic-driven health risks across urban water systems. A systematic literature review screened 1,742 records, of which 47 studies met the predefined eligibility criteria and were included in the final analysis. Here, we integrate machine learning, Bayesian inference, and structural equation modeling (SEM) to resolve microplastic contamination dynamics and associated health impacts across urban water systems. The synthesized dataset incorporated environmental source characteristics, contamination indicators, exposure metrics, biological response proxies, and health-related outcomes to support integrated modeling and risk prediction. Random Forest clustering identified six distinct contamination–health regimes (R<sup>2</sup> = 0.73), including industrial discharge-dominated, wastewater-associated, stormwater-influenced, mixed urban residential, landfill/leachate-associated, and low-contamination background conditions. Random Forest regression predicted health impairment with high accuracy (R<sup>2</sup> = 0.84; RMSE = 3.77). Additionally, microplastic health risk is governed by coupled environmental pressures, exposure pathways, and biological responses rather than isolated factors. Binary classification achieved 80% accuracy for detecting regulatory exceedance risk. Bayesian correlation revealed strong coupling between microplastics and exposure, oxidative stress, and inflammation. Structural equation modeling demonstrated that land-use intensity impacts health both directly and indirectly through wastewater loading and biological stress pathways, explaining 89.9% of health variance. This study integrates systematic evidence synthesis, machine learning, Bayesian inference, within a unified predictive framework, providing a novel approach for quantifying contamination–health relationships and supporting evidence-based urban water management. These findings establish a data-driven framework for predicting microplastic risks and optimizing mitigation strategies in urban water cycles.</p> Graphical abstract <p></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

How microplastics influence health risk pathways in urban water systems? integrated binary classification, machine learning and structural equation modeling approach

  • 期刊论文

  • 发表日期：

    2026年7月30日

查看全文
  • Hamid Rehman,
  • Bareera Maryem,
  • Eyup Debik,
  • Neslihan Manav-Demir,
  • Hanife Sari Erkan,
  • Mahdi Hassan,
  • Guleda Onkal Engin,
  • Kubra Ulucan-Altuntas,
  • Ahsanullah Soomro,
  • Mazhar Iqbal,
  • Sanjay K. Mohanty

摘要

Microplastic contamination in urban water systems poses emerging risks to human health, yet integrated frameworks linking sources, exposure pathways, and health outcomes remain limited. This study develops an integrated machine learning–Bayesian–structural equation modeling framework to quantify and predict microplastic-driven health risks across urban water systems. A systematic literature review screened 1,742 records, of which 47 studies met the predefined eligibility criteria and were included in the final analysis. Here, we integrate machine learning, Bayesian inference, and structural equation modeling (SEM) to resolve microplastic contamination dynamics and associated health impacts across urban water systems. The synthesized dataset incorporated environmental source characteristics, contamination indicators, exposure metrics, biological response proxies, and health-related outcomes to support integrated modeling and risk prediction. Random Forest clustering identified six distinct contamination–health regimes (R2 = 0.73), including industrial discharge-dominated, wastewater-associated, stormwater-influenced, mixed urban residential, landfill/leachate-associated, and low-contamination background conditions. Random Forest regression predicted health impairment with high accuracy (R2 = 0.84; RMSE = 3.77). Additionally, microplastic health risk is governed by coupled environmental pressures, exposure pathways, and biological responses rather than isolated factors. Binary classification achieved 80% accuracy for detecting regulatory exceedance risk. Bayesian correlation revealed strong coupling between microplastics and exposure, oxidative stress, and inflammation. Structural equation modeling demonstrated that land-use intensity impacts health both directly and indirectly through wastewater loading and biological stress pathways, explaining 89.9% of health variance. This study integrates systematic evidence synthesis, machine learning, Bayesian inference, within a unified predictive framework, providing a novel approach for quantifying contamination–health relationships and supporting evidence-based urban water management. These findings establish a data-driven framework for predicting microplastic risks and optimizing mitigation strategies in urban water cycles.

Graphical abstract