Table 1.
Examples of studies focusing on applying ML algorithms on patients’ clinical features of COVID-19
| Source | Objective | Dataset size | Features | Algorithms | Accuracy |
|---|---|---|---|---|---|
| [10] | Proposing and validating a diagnostic model for COVID-19 based on clinical and radiological features | 136 (COVID-19 patients (N = 70) and non-COVID-19 pneumonia patients (N = 66)) | 67 features (41 images + 26 clinical) |
C Model R Model CR Model |
95.2% 96.9% 98.6% |
| [8] | Evaluating clinical and imaging features for measuring the need for intensive care unit (ICU) treatment | 65 | Clinical, laboratory, and imaging features | Multivariate random forest modeling | 80% |
| [7] | Identifying the positive COVID-19 cases based on blood tests analysis | 279 | Patient’s age, gender, blood tests, and RT-PCR tests for COVID-19 |
Decision Tree Three-Way Random Forest (TWRF) classifier |
82% − 86% |
| [5] | Identifying the positive COVID-19 cases based on blood tests analysis | 786 | 81 COVID-19 (+), 517 COVID-19 (-), and 188 Pathogens (non COVID-19) | ANN classifier | 90% |
| [25] | Chest CT image-based-diagnose of COVID-19 | 275 | 88 COVID-19 (+) Chest CT images, 101 Bacterial Pneumonia (+) Chest CT images, and 86 Chest CT images of healthy people | DeepPneumonia | 99% |
| [4] | Chest CT image-based-diagnose of COVID-19 | 1020 | CT images (50% of COVID-19 patients) | 10 Convolutional Neural Networks: AlexNet, VGG-16, VGG-19, SqueezeNet, GoogleNet, MobileNet-V2, ResNet-18, ResNet-50, ResNet-101, and Xception | 99.51% |
| [16] | Chest X-Ray image-based-diagnose of COVID-19 | 1157 | 157 Pneumonia (+) Chest X-Ray images, 500 Pneumonia (+) Chest X-Ray images, and 500 Chest X-Ray images of healthy people | CoroNet | 90.21% |
| [13] | Chest CT image-based-diagnose of COVID-19 | 460 | 230 CT images from 79 COVID-19 patients, 100 CT images from 100 common pneumonia patients, and 130 CT images from 130 healthy people | AD3D-MIL | 97.9% |
| [26] | Chest X-Ray image-based-diagnose of COVID-19 | 3150 | 1050 COVID-19 (+) Chest X-Ray images, 1050 no-findings Chest X-Ray images, and 1050 pneumonia Chest X-Ray images | Capsule networks |
84.22% (multi-class) 97.24% (binary-class) |
| [24] | Chest X-Ray image-based-diagnose of COVID-19 | 381 | 127 COVID-19 (+) Chest X-Ray images and 127 Pneumonia (+) Chest X-Ray images | ResNet50 plus SVM | 95.33% |
| [21] | Chest X-Ray image-based-diagnose of COVID-19 | 16,700 | 313 COVID-19 (+) Chest X-Ray images, 2780 Bacterial Pneumonia (+) Chest X-Ray images, 6012 unknown Pneumonia Chest X-Ray images, and 7595 Chest X-Ray images of healthy people |
Weighted averaging (iteratively pruned) |
99.01% |