Predicting the Peak and COVID-19 trend in six high incidence countries: A study based on Modified SEIRD model

Abstract

The novel Coronavirus (COVID-19) has claimed the lives of almost a million people across the globe and this trend continues to rise rapidly day by day. The fear of getting infected by Corona virus is affecting the people emotionally, psychologically and mentally. They are not able to work to their full capacity and are also worried about the well beings of their near and dear ones. The National governments have taken up several measures like lockdowns, closing of educational institutions, and work from home for employees of companies wherever feasible. Governments are also advising people to take precautions like not to go out if not necessary, use of mask and keep a distance of appx. 6 ft. if you need to go out as the virus spreads from human to human in close proximity. These measures have helped to limit the spread of this virus in the past few months. However, due to rapid increase in the daily confirmed cases, it is becoming tougher for the governments to provide facilities like testing kits, hospitalization facilities, oxygen cylinders etc. to the infected persons. Thus, there is an urgent need to accurately estimate the number of cases in coming future that can help governments in acquiring the required resources. Further, to handle the economic distress caused by this virus, long-term planning is equally important. Focusing on these two aspects, this paper proposes to use the Modified SEIRD (Susceptible-Exposed-Infected-Recovered-Deceased) model to predict the peak and spread trend of COVID-19 in six countries namely USA, India, Brazil, Russia, Peru and Colombia having the highest number of confirmed cases. As in COVID-19, even infected asymptomatic persons can spread the infection, the chosen model is well suited as exposed compartment of SEIRD model includes asymptomatic exposed individuals which are infectious. Epidemiological data till 9th September 2020 has been utilised to perform short-term predictions till 31st December 2020. Long-term predictions have been computed till 31st December 2023, to estimate the end of the virus in the above-mentioned six nations. Small values of MAPE (Mean Absolute Percentage Error) have been obtained for the models fitted to reported data for all the countries. Student t-test has been used for accepting the predictions of the Modified SEIRD model based on the reported data.

The evaluation of COVID-19 prediction precision with a Lyapunov-like exponent

“…For the task, various types of prediction models have been used, such as compartmental models including SIR, SEIR, SEIRD and other modifications, see e.g. [8][9][10][11][12], artificial neural network models [13][14][15][16], Gompertz and logistics functions [17][18][19], ARIMA models [13,20], and many other approaches, see e. g. [21][22][23][24]. A survey on how deep learning and machine learning is used for COVID-19 forecasts can be found e.g.…”