Article Preview
TopIntroduction
In December, 2019, the first coronavirus disease 2019 (COVID-19) case was detected in the city of Wuhan, China (Cortegiani, Ingoglia, Ippolito, Giarratano, & Einav, 2020; Khan & Atangana, 2020; Shereen, Khan, Kazmi, Bashir, & Siddique, 2020). The COVID-19 is a highly contagious infection caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV2) (Shereen et al., 2020; Rothan & Byrareddy, 2020), which symptoms include fever, coughing and problems with breathing. According to the report by the World Health Organization (WHO), as of 29th June, 2020, a total of 10,021,401 COVID-19 cases were confirmed across the world (WHO, 2020). Out of the total confirmed cases, a total of 499,913 deaths were recorded from the viral infection from inception of the pandemic to 29th June, 2020.
Literature evidence suggests that artificial intelligence (AI) is a valuable tool in the efforts to contain the pandemic. Vaishya, Javid, Khan, and Haleem (2020) identified several ways in which AI can be applied in the fight against COVID-19. According to Vaishya et al (2020), AI has capabilities to detect early infection of the virus in an individual, monitoring of COVID-19 patient’s condition, and contact tracing of exposed persons. Mei et al. (2020) deployed AI to develop a rapid system for diagnosing COVID-19 patients. The diagnostic system uses AI algorithms to integrate chest computed tomography (CT) findings with exposure history, laboratory testing and clinical symptoms to diagnose COVID-19 positive persons. Among other merits of the AI-enabled diagnostic system such as enhanced accuracy, the system is reported to produce test results faster than the regular virus-specific reverse transcriptase polymerase chain reaction test, which takes two days to complete (Mei et al, 2020). In another study, Muhammad, Islam, Usman, and Ayon (2020) investigated the best classification algorithm for use in constructing a model for predicting the possibility of recovery from COVID-19 infection by patients. The experiments were conducted using COVID-19 data in South Korea and the predictive accuracies of support vector machines, decision tree, logistic regression, naïve Bayes, random forest, and k-nearest neighbor models were noted. The study concluded that the decision tree, which achieved 99.85% predictive accuracy, is the best classification algorithm for use in the prediction of whether or not a patient is going to recover from the infection (Muhammad et al., 2020). The research by Tuli Shreshth, Tuli Shikhar, Tuli Rakesh, and Gill (2020) integrated machine learning with cloud computing to develop a model that can predict the growth and trend of COVID-19 across countries of the world. The study used the iterative weighting in fitting a generalized inverse Weibull distribution, which served the purpose of developing a predictive model. The prediction framework, which can be deployed on a cloud computing platform, is capable to perform real-time prediction of the growth pattern and trend of the epidemic across the world (Tuli Shreshth et al., 2020).
One of the effective ways of controlling the spread of coronavirus is to determine the prevalent transmission mode from person-to-person within a geographic area. Knowledge of the dominant transmission mode specific to an area is vital so that experts can advise on strategies to be put in place to tame the spread. Against this backdrop, this study was aimed at determining the most appropriate classification algorithm to be adopted for predicting the prevalent transmission mode of coronavirus in a given area. Several classification algorithms are in existence, but not all of these can be deployed in every problem domain. The question of which machine learning algorithm will produce optimum results has to be determined experimentally through spot-checking. In the end of this study, the classification algorithm that offers the highest performance in predicting the dominant COVID-19 transmission mode within an area is advised.