Implementation of stacking based ARIMA model for prediction of Covid-19 cases in India

Abstract

Background Time-series forecasting has a critical role during pandemics as it provides essential information that can lead to abstaining from the spread of the disease. The novel coronavirus disease, COVID-19, is spreading rapidly all over the world. The countries with dense populations, in particular, such as India, await imminent risk in tackling the epidemic. Different forecasting models are being used to predict future cases of COVID-19. The predicament for most of them is that they are not able to capture both the linear and nonlinear features of the data solely. Methods We propose an ensemble model integrating an autoregressive integrated moving average model (ARIMA) and a nonlinear autoregressive neural network (NAR). ARIMA models are used to extract the linear correlations and the NAR neural network for modeling the residuals of ARIMA containing nonlinear components of the data. Comparison: Single ARIMA model, ARIMA-NAR model and few other existing models which have been applied on the COVID-19 data in different countries are compared based on performance evaluation parameters. Result The hybrid combination displayed significant reduction in RMSE(16.23%), MAE(37.89%) and MAPE (39.53%) values when compared with single ARIMA model for daily observed cases. Similar results with reduced error percentages were found for daily reported deaths and cases of recovery as well. RMSE value of our hybrid model was lesser in comparison to other models used for forecasting COVID-19 in different countries. Conclusion Results suggested the effectiveness of the new hybrid model over a single ARIMA model in capturing the linear as well as nonlinear patterns of the COVID-19 data.

Forecasting COVID-19 new cases using deep learning methods

“…Swaraj et al developed a model integrating ARIMA and nonlinear autoregressive neural network (NAR) for predicting the COVID-19 outbreak in India. The result shows a significant reduction in evaluation metrics (RMSE: 16.23%, MAE: 37.89% and MAPE: 39.53%) with the hybrid model compared to the single ARIMA model [ 19 ]. Besides, Wadhwa et al studied the effects of lockdown policy on disease transmission by predicting the number of active cases all over India.…”

Novel informatics approaches to COVID-19 Research: From methods to applications

“…Methods for short-term forecasting are usually data-driven, but they can also incorporate domain knowledge in the form of compartmental modelling (i.e., SIR-type models – see [48] ), as a means of integrating and simultaneously forecasting in a coherent manner multiple epidemiological indicators, such as cases, hospitalizations, and deaths. The study by Swaraj et al [21] uses a statistical data-driven approach, Aljaaf et al [22] , Dairi et al [23] , and Safari et al [24] use a machine learning data-driven approach, while Jing et al [25] uses compartmental modelling. Most of the forecasting studies were conducted as stand-alone evaluations or comparisons of methods by researchers who were not directly affiliated with public health authorities mandated to perform surveillance.…”

“…Most of the forecasting studies were conducted as stand-alone evaluations or comparisons of methods by researchers who were not directly affiliated with public health authorities mandated to perform surveillance. In terms of epidemiological indicators being forecast, the studies focused on confirmed cases, hospitalizations and deaths, although the contribution of smartphone mapping data, was explored by Jing et al [25] Methodologically, most studies used statistical and ML methods, either in comparison (Dairi et al [23] and Safari et al [24] ), within an ensemble (Aljaaf et al [22] and Swaraj et al [21] ), or as a part of a pipeline. Although results varied, findings were consistent with a recent review of methods for aberration detection in public health, with ML methods tending to outperform statistical methods for short-term forecasting.…”

“… Special Communication [17] ConceptWAS: A high-throughput method for early identification of COVID-19 presenting symptoms and characteristics from clinical notes Zhao, J. Special Communication [18] A retrospective approach to evaluating potential adverse outcomes associated with delay of procedures for cardiovascular and cancer-related diagnoses in the context of COVID-19 Zheng, N. S. Original Research [19] Learning from past respiratory failure patients to triage COVID-19 patient ventilator needs: A multi-institutional study Carmichael, H. Special Communication [20] A predictive model of clinical deterioration among hospitalized COVID-19 patients by harnessing hospital course trajectories Mauer, E. Original Research Forecasting and epidemic modelling (9) [21] Implementation of stacking based ARIMA model for prediction of Covid-19 cases in India Swaraj, A. Original Research [22] A fusion of data science and feed-forward neural network-based modelling of COVID-19 outbreak forecasting in IRAQ Aljaaf, A. J.…”

Suburban Road Networks to Explore COVID-19 Vulnerability and Severity

“…Similar to effective forecasting models for other epidemics, Hernandez-Matamoros [ 18 ] assessed the autoregressive integrated moving average (ARIMA) model for 145 countries in six regions, and this turned out to be also effective for COVID-19. Swaraj [ 19 ] proposed an ARIMA-based model that could capture the linear and non-linear components of the data by integrating an autoregressive neural network (NAR). Machine-learning and deep-learning models have also shown outstanding ability to forecast from time-series data, especially Long Short-Term Memory (LSTM), due to their capability to unveil dependencies over a long distance in time.…”