Table 1.
Results for screening models.
| Study, Country, Outcome | No. of CPP* | AI methods | Predictors | Val. methods | Performance (AUC, Accuracy (Acc%), Sensitivity (SEN%), Specificity (SPE%), PPV/NPV (%), (95% CI)) | Risk of Bias**: Participants/Predictors/Outcome/Analysis/Overall | ||||
|---|---|---|---|---|---|---|---|---|---|---|
| Yang et al. [4], USA, Early and rapid identification of high-risk SARS-CoV-2 infected patients | 1,898 | LR, DT, RF, GBDT | age, gender, race and 27 routine laboratory tests | 5-FCV | AUC 0.854 (95% CI: 0.829–0.878) | L | U | H | H | H |
| Li et al. [63], China, Screening based on ocular surface features | 104 | DL | Imaging features | 5-FCV | AUC 0.999 (95%CI, 1670.997–1.000, SEN 98.2, SPE 97.8 | U | U | U | H | H |
| AS Soltan et al. [3], UK, Early detection, Screening | 437 | multivariate LR, RF, XGBoost | Presentation laboratory tests and vital signs | TTS, 10-FCV | ED model: AUC 0.939, SEN 77.4, SPE 95.7Admissions model: AUC 0.940, SEN 77.4, SPE 94.8Both models achieve high NPP (>99) | H | H | H | H | H |
| Nan et al. [57], China, Early screening | 293 | DL, LR, SVM, DT, RF | 4 epidemiological features, 6 clinical manifestations (muscle soreness, dyspnea, fatigue, lymphocyte count, WBC, imaging features) | TTS | AUC 0.971, Acc 90, SPE 0.95 (LR optimal screening model) | H | U | H | H | H |
| Soares et al. [58], Brazil, Screening of suspect COVID-19 patients | 81 | ML, SVM, SMOTE Boost, ensembling, k-NN | Hemogram: (Red blood cells, MCV, MCHC, MCH, RDW, Leukocytes, Basophils, Monocytes, Lymphocytes, Platelets, Mean platelet volume, Creatinine, Potassium, Sodium, CRP, Age | unspecified | AUC 86.78 (95%CI: 85.65–87.90), SEN 70.25 (95%CI: 66.57–73.12), SPE 85.98 (95%CI: 84.94–86.84), NPV 94.92 (95%CI: 94.37–95.37), PPV 44.96 (95%CI: 43.15–46.87) | L | U | H | H | H |
| Feng et al. [59], China, Early identification of suspected COVID-19 pneumonia on admission | 32 | ML, LR (LASSO), DT, Adaboost | lymphopenia, elevated CRP and elevated IL-6 on admission | 10-FCV | AUC 0.841, SPE 72.7 | H | H | H | H | H |
| Wu et al. [60], China, Early detection | 27 | RF | 11 key blood indices: TP, GLU, Ca, CK-MB, Mg, BA, TBIL, CREA, LDH, K, PDW | 10-FVC, Ext. Val. | Acc 95.95, SEN 95.12, SPE 96.97 | L | L | L | H | H |
| Banerjee et al. [61], Brazil, Initial screening | 81 | RF, ANN | platelets, leukocytes, eosinophils, basophils, lymphocytes, monocytes. | 10-FCV | AUC 0.95 | H | H | H | H | H |
| Peng et al. [62], China, Quick and accurate diagnosis | 32 | SRLSR, non-dominated radial slots-based algorithm, ARMED, GFS, RFE | 18 diagnostic factors: WBC, eosinophil count, eosinophil ratio, 2019 new Coronavirus RNA (2019n-CoV), Amyloid-A, Neutrophil ratio, basophil ratio, platelet, thrombocytocrit, monocyte count, procalcitonin, neutrophil count, lymphocyte ratio, lymphocyte count, monocyte ratio, MCHC, Urine SG | not performed | not performed | L | L | U | H | H |
*CPP = COVID-19 Positive Patients, Abbreviations of medical terms included in this Table are provided in the Appendix.
**L: Low, H: High, U: Unclear