Comparative Evaluation of the Multilayer Perceptron Approach with Conventional ARIMA in Modeling and Prediction of COVID-19 Daily Death Cases

Abstract

COVID-19 continues to pose a dangerous global health threat, as cases grow rapidly and deaths increase day by day. This increasing phenomenon does not only affect economic policy but also international policy around the world. In this paper, Pakistan daily death cases of COVID-19, from February 25, 2020, to March 23, 2022, have been modeled using the long-established autoregressive-integrated moving average (ARIMA) model and the machine learning multilayer perceptron (MLP) model. The most befitting model is selected based on the root mean square error (RMSE), mean square error (MSE), and mean absolute error (MAE). Values of the key performance indicator (KPI) showed that the MLP model outperformed the ARIMA model. The MLP model with 20 hidden layers, which emerged as the overall most apt model, was used to predict future daily COVID-19 deaths in Pakistan to enable policymakers and health professionals to put in place systematic measures to reduce death cases. We encourage the Government of Pakistan to intensify its vaccination campaign and encourage everyone to get vaccinated.

Comparative effectiveness analysis of univariate time-series forecasting models for disease mortality rates in the global burden of disease database: a case study of global hypertensive heart disease among women of childbearing age

“…This equivalence was consistently observed across three distinct loss functions (MSE, MAE, and MAPE), each penalizing different types of forecast errors, which underscores the robustness of their performance. Previous studies in infectious disease forecasting ( 13 , 14 ) have conclusively demonstrated the superiority of nonlinear models over traditional ARIMA, a conclusion our research strongly supports for chronic disease burden data. However, many existing studies in economics ( 7 , 8 ) and environmental science ( 9 , 11 ) tend to focus on declaring a single “best” model or on building complex ensembles.…”

Comparative effectiveness analysis of univariate time-series forecasting models for disease mortality rates in the global burden of disease database: a case study of global hypertensive heart disease among women of childbearing age

“…This equivalence was consistently observed across three distinct loss functions (MSE, MAE, and MAPE), each penalizing different types of forecast errors, which underscores the robustness of their performance. Previous studies in infectious disease forecasting ( 13 , 14 ) have conclusively demonstrated the superiority of nonlinear models over traditional ARIMA, a conclusion our research strongly supports for chronic disease burden data. However, many existing studies in economics ( 7 , 8 ) and environmental science ( 9 , 11 ) tend to focus on declaring a single “best” model or on building complex ensembles.…”

Respiratory viral infections: when and where? A scoping review of spatiotemporal methods

“…The Prophet model [ 22 - 25 ] has been widely used to predict potential COVID-19 outbreaks. Other ML models aiding in disease trend predictions include the multilayer perceptron [ 26 - 29 ], support vector machine [ 25 , 26 , 30 - 33 ], tree-based models including random forest [ 28 , 30 , 33 - 37 ], gradient boosting [ 34 ], XGBoost [ 32 , 35 ], and regularised regression models [ 38 ], including Lasso and Elastic Net. Deep learning [ 39 ], a subset of ML, excels at processing large data sets to identify intricate patterns, yielding accurate forecasts.…”

“…ARIMA and linear regression) often underfit large-scale epidemic data and fail to capture intricate spatiotemporal patterns. Although DL architectures, such as transformers [ 88 ], excel at processing large-scale data, they can overfit or become unstable when the data are sparse or imbalanced [ 29 ]. Rigorous evaluation of model generalisability via cross-validation, holdout testing, and bootstrapping is therefore essential to select optimal models according to the data set's characteristics.…”

“…Third, inadequate data segmentation. Qureshi et al [ 29 ], applied year-round data, without segmenting data based on its characteristics, to develop tailored models. Additionally, neglect spatial dependencies in multi-regional forecasting.…”

Predicting factors for extremity fracture among border-fall patients using machine learning computing

“…MPNN was selected in this study because it is one of the most used neural network architectures. MPNN makes good classifier algorithms and has been used in medical research [ [18] , [19] , [20] ].…”

Comparative effectiveness analysis of univariate time-series forecasting models for disease mortality rates in the global burden of disease database: a case study of global hypertensive heart disease among women of childbearing age

“…This equivalence was consistently observed across three distinct loss functions (MSE, MAE, and MAPE), each penalizing different types of forecast errors, which underscores the robustness of their performance. Previous studies in infectious disease forecasting ( 13 , 14 ) have conclusively demonstrated the superiority of nonlinear models over traditional ARIMA, a conclusion our research strongly supports for chronic disease burden data. However, many existing studies in economics ( 7 , 8 ) and environmental science ( 9 , 11 ) tend to focus on declaring a single “best” model or on building complex ensembles.…”

Respiratory viral infections: when and where? A scoping review of spatiotemporal methods

“…The Prophet model [ 22 - 25 ] has been widely used to predict potential COVID-19 outbreaks. Other ML models aiding in disease trend predictions include the multilayer perceptron [ 26 - 29 ], support vector machine [ 25 , 26 , 30 - 33 ], tree-based models including random forest [ 28 , 30 , 33 - 37 ], gradient boosting [ 34 ], XGBoost [ 32 , 35 ], and regularised regression models [ 38 ], including Lasso and Elastic Net. Deep learning [ 39 ], a subset of ML, excels at processing large data sets to identify intricate patterns, yielding accurate forecasts.…”

“…ARIMA and linear regression) often underfit large-scale epidemic data and fail to capture intricate spatiotemporal patterns. Although DL architectures, such as transformers [ 88 ], excel at processing large-scale data, they can overfit or become unstable when the data are sparse or imbalanced [ 29 ]. Rigorous evaluation of model generalisability via cross-validation, holdout testing, and bootstrapping is therefore essential to select optimal models according to the data set's characteristics.…”

“…Third, inadequate data segmentation. Qureshi et al [ 29 ], applied year-round data, without segmenting data based on its characteristics, to develop tailored models. Additionally, neglect spatial dependencies in multi-regional forecasting.…”

Predicting factors for extremity fracture among border-fall patients using machine learning computing

“…MPNN was selected in this study because it is one of the most used neural network architectures. MPNN makes good classifier algorithms and has been used in medical research [ [18] , [19] , [20] ].…”

Comparative effectiveness analysis of univariate time-series forecasting models for disease mortality rates in the global burden of disease database: a case study of global hypertensive heart disease among women of childbearing age

“…This equivalence was consistently observed across three distinct loss functions (MSE, MAE, and MAPE), each penalizing different types of forecast errors, which underscores the robustness of their performance. Previous studies in infectious disease forecasting ( 13 , 14 ) have conclusively demonstrated the superiority of nonlinear models over traditional ARIMA, a conclusion our research strongly supports for chronic disease burden data. However, many existing studies in economics ( 7 , 8 ) and environmental science ( 9 , 11 ) tend to focus on declaring a single “best” model or on building complex ensembles.…”

Respiratory viral infections: when and where? A scoping review of spatiotemporal methods

“…The Prophet model [ 22 - 25 ] has been widely used to predict potential COVID-19 outbreaks. Other ML models aiding in disease trend predictions include the multilayer perceptron [ 26 - 29 ], support vector machine [ 25 , 26 , 30 - 33 ], tree-based models including random forest [ 28 , 30 , 33 - 37 ], gradient boosting [ 34 ], XGBoost [ 32 , 35 ], and regularised regression models [ 38 ], including Lasso and Elastic Net. Deep learning [ 39 ], a subset of ML, excels at processing large data sets to identify intricate patterns, yielding accurate forecasts.…”

“…ARIMA and linear regression) often underfit large-scale epidemic data and fail to capture intricate spatiotemporal patterns. Although DL architectures, such as transformers [ 88 ], excel at processing large-scale data, they can overfit or become unstable when the data are sparse or imbalanced [ 29 ]. Rigorous evaluation of model generalisability via cross-validation, holdout testing, and bootstrapping is therefore essential to select optimal models according to the data set's characteristics.…”

“…Third, inadequate data segmentation. Qureshi et al [ 29 ], applied year-round data, without segmenting data based on its characteristics, to develop tailored models. Additionally, neglect spatial dependencies in multi-regional forecasting.…”

Predicting factors for extremity fracture among border-fall patients using machine learning computing

“…MPNN was selected in this study because it is one of the most used neural network architectures. MPNN makes good classifier algorithms and has been used in medical research [ [18] , [19] , [20] ].…”