Abstract
The improper disposal of COVID-19-related waste, such as masks, gloves, and personal protective equipment, has contributed to the release of nano-plastics, which degrade into nano-sized particles, posing severe environmental risks. Understanding the impacts of these nano-level particles is essential for mitigating long-term ecological harm. A comparative analysis was conducted to evaluate the effectiveness of the deep multi-layer perceptron (DMLP) algorithm in predicting the environmental impacts of pollution from nano-level plastic pollution from COVID-19 waste. The study focused on the degradation and dispersion of nano-plastics in Isfahan, Iran. The DMLP’s performance was benchmarked against multiple classifiers to assess its ability to capture intricate patterns and high-level feature representations unique to nano-plastic behavior. Hyperparameter optimization played a central role, fine-tuning learning rates, activation functions, and network configurations to enhance the DMLP’s predictive capabilities. The model’s performance was rigorously evaluated using metrics such as accuracy and the area under the receiver operating characteristic curve (AUC-ROC). The analysis specifically targeted nano-plastic dispersal patterns, identifying hot spots and ecological risks, to provide a comprehensive understanding of COVID-19 waste-related environmental impacts. The DMLP algorithm demonstrated exceptional performance in modeling and predicting the environmental dispersion of nano-plastics originating from COVID-19 waste in Isfahan, Iran. By leveraging its deep architecture, the DMLP effectively identified complex patterns in the degradation and movement of nano-plastics, outperforming other classifiers in accuracy and AUC-ROC metrics. Innovative visualization methods were employed to interpret the model’s decision boundaries, providing clear insights into high-risk zones and potential environmental impacts. The findings revealed critical nano-plastic accumulation hot spots, particularly in urban and water-sensitive areas, underscoring the vulnerability of these ecosystems to pollution from COVID-19 waste. This study demonstrates the capability of the DMLP to uncover intricate relationships within environmental datasets, particularly in analyzing the impacts of nano-level COVID-19 waste. The model’s ability to accurately predict the dispersion and accumulation of nano-plastics provides valuable insights into high-risk ecological zones, underscoring its potential to support environmental decision-making. These findings underscore the critical need for targeted waste management strategies to mitigate the long-term environmental impacts of nano-plastics and highlight the DMLP’s applicability in addressing complex pollution challenges in dynamic urban ecosystems.