Coronavirus disease (COVID-19) emerged in December 2019 and within three months, it was declared a global pandemic [1]. COVID-19 represents a diverse family of positive RNA virus capable of causing severe respiratory disease in humans and animals [2]. COVID-19 is a new member of beta coronavirus and is related to severe acute respiratory syndrome coronavirus (SARS-CoV) [3]. SARS-CoV emerged in 2002 and spread to 26 countries, infecting more than 8000 patients. SARS-CoV was eventually contained through public health measures, as there was no approved vaccine for SARS-CoV infections. With the second phase of a smaller outbreak in 2004, coronavirus has since not affected humans to that extent after the initial outbreak ended. However, SARS-CoV-like virus continued to circulate in bats and to date, not much effort has been made to treat and control coronavirus infections.
A coronavirus outbreak is a threat to the global economy and health security. The outbreak is likely to continue for months as there is no treatment currently available to prevent COVID-19 infections. Compared to SARS-CoV, COVID-19 exhibits faster human-to-human transmission, leading to the declaration of a global pandemic with millions of infections and thousands of deaths worldwide. The rapid spread of COVID-19 disease and its impact on the economy has emphasized the development of coronavirus vaccines. To this end, the research on COVID-19 global pandemic and future growth is a hot topic, and many studies are being conducted on the preventive measures of COVID-19. On the other hand, the rush towards vaccine development is in progress and may take many months for its validation and test. With the current situation of lockdown, infections, and deaths, it is crucial to know when the spreading of COVID-19 will be over.
With the current number of COVID-19 cases for each country, the trend shows that the cases per day is still increasing for a number of the countries, and it is not clear by when the spread of COVID-19 can be contained. If the dates by when COVID-19 can be contained is known for each country, it will help the government, policymakers, entrepreneurs and businesses make appropriate decisions that could potentially impact public and social matters. To address the above concerns, computational techniques [4,5,6] can be applied to model the spread of COVID-19 and forecast the dates by when a country might be able to contain the spread of COVID-19. In [6], the authors have used the fractional-order susceptible individuals, asymptomatic infected, symptomatic infected, recovered, and deceased (SEIRD) model for forecasting the spread of COVID-19. Using Italy’s data, they showed that the fractional-order model has lower root-mean-square error (RMSE) than the classical one. The authors in [5] proposed a autoregressive integrated moving average (ARIMA) model for forecasting the expected number of daily new cases of COVID-19 in Saudi Arabia. They forecasted the COVID-19 cases for the next 4 weeks. Four different models were evaluated namely the autoregressive model, moving average, a combination of both autoregressive moving average and ARIMA. They found out that the ARIMA model performed well using Saudi Arabia data. Models from the exponential smoothing family is used for forecasting the cumulative daily confirmed cases, deaths and recoveries globally in [4]. Exponential smoothing models is used as the authors claim that it can capture a variety of trend, seasonal forecasting patterns and is mostly suitable for short time series data. An autoregressive neural network approach has been proposed by Saba and Elsheikh [7] for the prediction of the prevalence of COVID-19 outbreak in Egypt. Compared to the officially reported cases, a good performance was noted for their approach. One of the studies [8] also revealed a significant relationship between air pollution and COVID-19 infection. In [9], the authors proposed a long short-term memory (LSTM) network for forecasting the air quality in ten different horizons in Delhi, India. Several approaches [10,11,12,13,14] have been recently proposed using deep learning methods for forecasting the spread of COVID-19. In [11], Chimmula and Zhand proposed an LSTM model for forecasting the number of COVID-19 cases in Canada. They predicted that Canada will be able to end the COVID-19 outbreak around June 2020. Their prediction was somewhat close as the number of cases was reducing from May 2020 before undergoing a second wave of COVID-19 outbreak. The transmission rate of Canada was also compared with that of the USA and Italy. Chandra, Jain and Chauhan [10] proposed three different approaches using LSTM, bidirectional LSTM and encoder-decoder LSTM models for forecasting the spread of COVID-19 infections among numerous selected states in India. Both univariate and multivariate models were considered. For multivariate model, the authors used input from the state that is considered and inputs from three adjacent states. The authors reported that the LSTM model performed well for most of the cases in comparison to the bidirectional LSTM and encoder-decoder LSTM models. A rolling mean of 3-days has been used for processing the data. Promising results were noted, and the approach can be used for forecasting the spread of COVID-19 infections in other countries or areas.
In this study, we obtain daily new cases and apply the long short-term memory (LSTM) architecture for forecasting the dates by when a country might be able to contain the spread of this novel coronavirus. The term ‘contain the spread of COVID-19’, refers that a country will be able to bring down its daily new cases of COVID-19 to less than 1% of its RMSE. We believe that the country will be able to control and manage the spread of any new cases of COVID-19 in future (if any) after it achieves cases less than 1% of its RMSE unless there are changes in travel and other restriction measures. The results obtained are promising and shows the forecasted number of new cases in the coming months for each country. The results also suggest that countries should not haste into easing restrictions as it can result in a second wave of COVID-19 outbreak in the country.