Classification of Covid-19 Patients by Clinical Symptoms and Demographic Characteristics Publisher



Shafiekhani S¹ ; Rezaei MA² ; Saberi M³ ; Raei M⁴ ; Mohammadi TC² Show Affiliations
Source: Journal of Military Medicine Published:2025

Abstract

Background and Aim: Effective screening can help alleviate the challenges faced by the healthcare and treatment system in diagnosing COVID-19. This study aimed to introduce a machine learning model for predicting the severity of COVID-19 based on symptom records, laboratory factors, and demographic characteristics. Methods: This study was conducted using checklists completed by individuals suffering from coronavirus at Baqiyatalah Hospital in Tehran, Iran. The correlation between data characteristics and the target variable (severity of coronavirus disease) was analyzed using MATLAB 2019b software. Various classification models, such as Support Vector Machine, Decision Tree, Naive Bayes, Ensemble, and K-Nearest Neighbor, were utilized to predict the severity of COVID-19 disease. Precision, accuracy, readability, F1 score, detectability, and the area under the ROC curve for different classification models were calculated as features scored by the MRMR (Maximum Relevance-Minimum Redundancy) algorithm. These models were trained with data from 480 subjects (80% of all subjects in the data collection) and tested with data from 120 subjects (20% of all subjects). The MRMR algorithm was used to score the features related to the severity of COVID-19 disease. Results: The results of the study showed that LDH, CABG, vertigo, MI, rheumatism, obsessive behaviors, recurrence of the disease, diabetes, gender, SGPT, hemoptysis, sleep problems, angiography, and dry cough were the top fifteen characteristics to predict the severity of the coronavirus disease, respectively. The Ensemble model was reported as the best model for detecting the severity of COVID-19 with the highest accuracy, the value of the area under the ROC curve, and the F1 score, 0.72, 0.71, and 0.76, respectively. Conclusion: This study presents a machine learning method to facilitate early clinical decision-making during the COVID-19 outbreak and a COVID-19 severity prediction model capable of efficiently screening and classifying individuals into two groups: severe and mild COVID-19 disease. © 2025 Baqiyatallah University of Medical Sciences. All rights reserved.