Related Experiment Video
Updated: Jul 27, 2025
Dynamic Monitoring of Seroconversion using a Multianalyte Immunobead Assay for Covid-19
Published on: February 16, 2022
A diagnostic model for COVID-19 based on proteomics analysis
Walaa Alkady1, Khaled ElBahnasy2, Walaa Gad2
1Bioinformatics Program, Faculty of Computer and Information Sciences, Ain Shams University, Cairo, Egypt.
Background And Objective:
Early diagnosis of Coronavirus Disease 2019 (COVID-19) can help save patients' lives before the disease turns severe. This can be achieved through an effective and correct treatment protocol. In this paper, a prediction model is proposed to detect infected cases and determine the severity level of the disease.
Methods:
The proposed model is based on utilizing proteins and metabolites as features for each patient, which are then analyzed using feature selection methods such as Principal Component Analysis (PCA), Information Gain (IG), and analysis of Variance (ANOVA) to select the most significant features. The model employs three classifiers, namely K-Nearest Neighbor (KNN), Support Vector Machine (SVM), and Random Forest (RF), to predict and classify the severity level of the COVID-19 infection. The proposed model is evaluated using four performance measures: accuracy, sensitivity, specificity, and precision.
Results:
The experiment results show that the proposed model accuracy can reach 80% using RF classifier with PCA. The PCA selects 22 proteins and 10 metabolites. While ANOVA selects 9 proteins and 5 metabolites. The accuracy reaches 92% after applying RF classifier with the ANOVA. Finally, the accuracy reaches 93% using the RF classifier with only ten features. The selected features are 7 proteins and 3 metabolites. Moreover, it shows that the selected features have a relation to the immune system and respiratory systems.
Conclusion:
The proposed model uses three classifiers and shows promising results by selecting the important features and maximizing the prediction accuracy.
