A ROUGH SET BASED CLASSIFICATION OF PANDAMIC PATIENTS: A CASE STUDY OF COVID-19

Abstract

COVID-19, caused by a novel strain of corona virus, has emerged as one of the most severe public health crises in modern history. Speedily growing across the world, it claimed lots of lives and infected another more. Like many viruses, the corona virus has undergone continuous mutations, producing variants such as Delta and Omicron that are highly infectious and responsible for large-scale fatalities in several countries. Although the rate of infection has declined in recent months, the pandemic has left a lasting impact on societies, economies, and everyday life. Until recently, governments and scientists worldwide faced the enormous challenge of containing the rapid transmission of the disease. Despite the wide availability of vaccines, the virus continues to evolve into stronger variants with higher transmission rates, as observed in recent surges in China and South Korea. At the beginning, symptoms were mostly mild fever, headache, and dry cough. Over time, however, a significant number of asymptomatic cases were reported, complicating detection and control measures. In the early stages, treatment approaches were largely symptom-based. As with other pandemics, the exponential growth of infections often led to acute shortages of medical resources. This situation was particularly visible during the first and second waves of COVID-19, when even technologically advanced nations with modern healthcare infrastructure struggled, resulting in high mortality. The key challenge in such scenarios is the optimal utilization of limited medical resources to reduce adverse outcomes. A promising way to address this issue is through the application of classification algorithms that can prioritize patients based on severity. In this article, we propose a rough set theory–based (RST) classification algorithm to categorize COVID-19 positive patients into predefined groups, enabling healthcare providers to manage each category more effectively. The proposed approach was validated by the experimental study on a real dataset.