Table 6.

Results for Prognostic Imaging 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
Fakhfakh et al. [42], Prognosis	42	RNN, CNN	Unclear	unspecified	Acc 92	H	U	U	L	H
Zhu et al. [66], China, Disease progression prediction	408	SVM, LR	Imaging features	5-FCV	Acc 85.91	L	U	U	H	H
Qi et al. [67], China, Hospital stay prediction (Short-term (<10 days), long-term (>10 days))	31	LR, RF	Imaging features (CT radiomics)	5-FCV	AUC 0.97, SEN 100, SPE 89, (95%CI 0.83–1.0)	U	L	L	H	H
Xiao et al. [68], China, Severity assessment, disease progression	408	DL, CNN, ResNet34 (RNN)	Imaging features	5-FCV	AUC 0.987 (95% CI: 0.968–1.00), Acc 97.4	L	U	U	H	H
Cohen et al. [36], Severity assessment for COVID19 Pneumonia	80	NN	CXR features	not performed		U	U	U	H	H
Salvatore et al. [69], Italy, Prognosis prediction (discharging at home, hospitalization in stable conditions, hospitalization in critical conditions, death)	98	LR	Imaging features	not performed	Acc 81, SEN 88, SPE 78	H	U	U	H	H
Liu et al. [44], China, Severity assessment	134	CNN	Imaging features (APACHE-II, NLR, d-dimer level)	Ext. val.	AUC 0.93, (95% CI: 0.87–0.99)	L	U	U	U	U

*CPP = COVID-19 Positive Patients, Abbreviations of medical terms included in this Table are provided in the Appendix.

**L: Low, H: High, U: Unclear