The predictive model for COVID-19 pandemic plastic pollution by using deep learning method

Area of Science:

• Environmental Science
• Public Health
• Computer Science

Background:

• The COVID-19 pandemic significantly increased single-use plastic waste, including personal protective equipment (PPE) and sanitizer bottles.
• This surge in pandemic plastics contributes to environmental pollution and poses risks for disease transmission.
• Effective waste management strategies require accurate prediction of plastic waste expansion.

Purpose of the Study:

• To develop and evaluate a deep learning-based predictive model for forecasting pandemic plastic waste in Iran's megacities.
• To assess the model's accuracy and compare its performance against traditional machine learning methods.
• To provide a tool for situational management and control procedures to mitigate environmental impacts and disease spread.

Main Methods:

• A comprehensive dataset was compiled, covering COVID-19 spread and PPE usage in Iran from February 2020 to October 2021.
• A deep neural network (DNN) model was trained using 80% of the data and validated on the remaining 20%.
• Model performance was rigorously evaluated using metrics like Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Mean Absolute Percentage Error (MAPE), and Receiver Operating Characteristic (ROC) curves, with comparisons to k-nearest neighbours, decision trees, random forests, support vector machines, Gaussian naïve Bayes, logistic regression, and multilayer perceptron.

Main Results:

• The DNN-based model demonstrated superior accuracy in predicting pandemic plastic pollution compared to other evaluated methods.
• The DNN model achieved a low error rate, with MSE = 0.024, RMSE = 0.027, and MAPE = 0.025.
• ROC curve analysis confirmed the DNN model's high performance, yielding an Area Under the Curve (AUC) of 0.929.

Conclusions:

• Deep learning, specifically the DNN model, offers a highly accurate and effective approach for forecasting pandemic plastic waste.
• The predictive capabilities of this model can significantly aid environmental management and public health interventions related to COVID-19 waste.
• Implementing such predictive models is crucial for developing targeted strategies to combat plastic pollution and its associated risks.