Covid-19 Patients Diagnosis (CPD) Strategy Using Data Mining Techniques

Mohamed, Alaa Mostafa; Saleh, Ahmed; A. Altantawy, Doaa; Abo-Elsoud, Mohy Eldin Ahmed

doi:10.21608/bfemu.2022.233811

Covid-19 Patients Diagnosis (CPD) Strategy Using Data Mining Techniques

Document Type : Research Studies

Authors

¹ Master Degree Researcher of Electronics and Communication Department, Faculty of Engineering, Mansoura University works at Delta Higher Institute for Engineering and Technology

² Professor at the Computers and Control Department, Faculty of Engineering, Mansoura University, Egypt.

³ Assistant Professor at the Electronics and Communication Department, Faculty of Engineering, Mansoura University, Egypt

⁴ Professor at the Electronics and Communication Department, Faculty of Engineering, Mansoura University, Egypt

10.21608/bfemu.2022.233811

Abstract

Covid-19, the world continues to live in anxiety and instability despite efforts to find a vaccine and emerge from this crisis. Especially after the emergence of a new mutated Corona virus called Omicron. This mutated sparked a state of controversy about the extent of its impact and its ability to spread among people. The Covid-19 epidemic has thrown the world economy into disarray. It also resulted in a widespread suspension of work and output throughout society, hurting economic and society. This paper introduces a Covid-19 Patients Diagnosis (CPD) strategy that works to find a fast and highly effective prognosis for diagnosing Covid-19 patients. The proposed strategy has two main stages named Feature Selection Stage (FSS) and Covid-19 Diagnosis Stage (CDS). The FSS has main objective to select the powerful features for the diagnosis stage. The features are selected in the FSS by using Chi-Square Feature Selection (CSFS) method. In fact, CSFS is a filter feature selection technique that has the ability to quickly choose the most effective subset of features. Then, quick and accurate diagnosis is provided by using Improved K-Nearest Neighbors (IKNN). The main idea in the proposed IKNN is that a circle with a radius value that equals the average distance of K of the closet items will be constructed and then the nearest M of items will be determined to classify the patient to the correct class “Covid” or “Non-Covid”. The results explain that the proposed strategy called CPD gives an accuracy of up to 96.36%.

Main Subjects