Performance evaluation of selected machine learning algorithms for COVID-19 prediction using routine clinical data:...

Area of Science:

• Medical Informatics
• Computational Biology
• Machine Learning Applications

Background:

• The diagnosis and treatment of coronavirus disease (COVID-19) remain complex due to its evolving nature.
• Early prediction frameworks are essential for managing COVID-19 effectively.
• Machine learning (ML) offers potential for extracting patterns from large datasets to build predictive models.

Purpose of the Study:

• To apply various machine learning techniques for developing clinical predictive models for COVID-19.
• To identify the best-performing ML model for COVID-19 diagnosis and screening.

Main Methods:

• Utilized a dataset of 501 case records (COVID-19 and non-COVID-19) with 32 diagnostic features from Ayatollah Talleghani hospital.
• Developed and evaluated ML algorithms including Naïve Bayesian, Bayesian Net, random forest (RF), multilayer perceptron, K-star, C4.5, and support vector machine.
• Assessed model performance using accuracy, sensitivity, specificity, precision, F-score, and ROC analysis.

Main Results:

• The random forest (RF) algorithm achieved the highest performance.
• RF demonstrated an accuracy of 92.42%, specificity of 75.70%, precision of 92.30%, sensitivity of 92.40%, F-measure of 92.00%, and ROC of 97.15%.

Conclusions:

• The RF model significantly outperformed six other classification models in COVID-19 prediction.
• Implementation of the RF model in healthcare settings can enhance diagnostic accuracy and speed for early prevention, screening, and treatment.