Prediction of Next Wave of Coronavirus Disease 2019 (COVID 19) Using Regression Model

Abstract

The Coronavirus Disease 2019 (COVID 19) has caused chaos everywhere in the world, and is prevailing with time. On 11th March 2020, World Health Organization (WHO) declared COVID-19 a global pandemic. Nonetheless, it is imperative to monitor the number of patients being affected. In this era of technological advancement and unprecedented change where the rate of innovations and discoveries are occurring at quite a pace, whilst in the field of health, a lot of work is still being done manually. Many advancements have been made towards the domain of treatment nevertheless, no such initiative has been taken to automate the process of making predictions of deadly viruses until the year 2020 when the world was suffering through a pandemic as last pandemic, Influenza occurred back in 1918. Predictions with respect to waves of virus were initiated in 2020 when WHO made the covid-19 data available for researchers. Every country is facing issues while handling coronavirus in terms of lack of vaccinations, in adequate medical facilities, shortage of oxygen and logistics. This due to sudden outbreaks of coronavirus that does not give ample time to countries to make necessary arrangements as there is no mechanism to predict the upcoming waves. To overcome this problem, in this research we have aimed to generalize the approach towards the prediction of the next wave using multiple linear regression model to help international community take necessary actions to contain the spread at global level. In our model total 17 attributes out of 113 were selected and 16 were independent variables (inputs) and 1 was a dependent variable (output). A study of trends of disease spread in the past and people affected by the disease daily are taken into consideration to predict the upcoming wave of COVID-19 globally. This will inform the region to investigate at country level to identify what Standard operating procedures are not being followed which will cause the predicted wave. Furthermore, prepare the countries of that region to arrange necessary medical equipment and take preventive measures to contain the spread in order to avoid the predicted wave. Moreover, this research fills many gaps highlighted during the previous researches such as handling of missing values in the input data given, skewed values, feature engineering and selection. For this work, dataset is taken from Our World in Data (OWID). The number of active cases of a certain period along with 15 other attributes are given as the input to the model and linear regression model techniques are applied to the data in order to predict the upcoming 6 COVID-19 wave based on the analysis performed on new active cases and deceased cases with respect to dates. Forecasting future trends will help international community to make necessary arrangement to contain the spread such as travel restrictions, focus of International community to expediating the process of vaccination and helping countries of a specific region which are unable to contain the spread. This pandemic is a global issue and apart from efforts being done by local governments of all countries, steps shall be taken on international platforms to eliminate this virus in a joint effort. The performance of the designed model is compared with the historic trends and other published methods, which demonstrates that the designed model provides predictions that are more accurate and precise up to 97.6%. The system is fast, efficient and has a high response rate as compared to the other models.