Scholarlink

<p>Covid-19 intensified the operational complexity and scale of medical waste collection and treatment. We study how circular economy-oriented recycling policies interact with sustainability indicators by integrating the best–worst method (BWM) for multi-criteria weighting with a deep learning (DL) model that predicts policy effectiveness. To ensure reproducibility, we provide full BWM matrices, consistency checks (CR = 0.006), and the DL architecture. In line with the scope, we articulate and evaluate the computational demands of large-scale policy simulation: our DL training and batched scenario evaluation are parallelizable and GPU-accelerated, and the workflow is designed to scale to national, multi-year datasets via data-parallel and model-parallel strategies. Results show that SDG 8 (Decent Work &amp; Economic Growth), SDG 17 (Partnerships), and SDG 6 (Clean Water &amp; Sanitation) receive the highest weights from experts, while SDG 3 (Good Health &amp; Well-Being) is ranked lower; we explain this counter-intuitive outcome via the economic leverage of recycling programs during crises. The DL classifier achieves 82.61% accuracy with balanced class metrics, and we report robustness checks. We discuss implications, limitations, and the scalability of our pipeline on high-performance computing (HPC) platforms for real-time decision support in future health emergencies. To enhance clarity, we explicitly describe the analytical pipeline: expert judgments inform the BWM-derived SDG weights, which are subsequently integrated into the deep learning classifier to evaluate scenario-level sustainability performance.</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

Integrated BWM-deep learning for medical waste collection in circular economy: sustainability considerations in pandemics

  • 期刊论文

  • 发表日期：

    2026年2月2日

查看全文
  • Samira Baratian,
  • Hamed Fazlollahtabar

摘要

Covid-19 intensified the operational complexity and scale of medical waste collection and treatment. We study how circular economy-oriented recycling policies interact with sustainability indicators by integrating the best–worst method (BWM) for multi-criteria weighting with a deep learning (DL) model that predicts policy effectiveness. To ensure reproducibility, we provide full BWM matrices, consistency checks (CR = 0.006), and the DL architecture. In line with the scope, we articulate and evaluate the computational demands of large-scale policy simulation: our DL training and batched scenario evaluation are parallelizable and GPU-accelerated, and the workflow is designed to scale to national, multi-year datasets via data-parallel and model-parallel strategies. Results show that SDG 8 (Decent Work & Economic Growth), SDG 17 (Partnerships), and SDG 6 (Clean Water & Sanitation) receive the highest weights from experts, while SDG 3 (Good Health & Well-Being) is ranked lower; we explain this counter-intuitive outcome via the economic leverage of recycling programs during crises. The DL classifier achieves 82.61% accuracy with balanced class metrics, and we report robustness checks. We discuss implications, limitations, and the scalability of our pipeline on high-performance computing (HPC) platforms for real-time decision support in future health emergencies. To enhance clarity, we explicitly describe the analytical pipeline: expert judgments inform the BWM-derived SDG weights, which are subsequently integrated into the deep learning classifier to evaluate scenario-level sustainability performance.