Table 2.

Review of algorithm approach for etiological diagnosis of uveitis and their diagnostic performance.

Study	Methods	Study Factors	Etiology	Diagnostic Performance
Gonzalez et al., 2016 [10]	Bayesian belief networks Dataset: 200 cases (anterior uveitis)	Demographic (gender) Ophthalmic (ocular symptoms and signs) Clinical examination (systemic symptoms and signs) Laboratory tests	11	Sensitivity (most probable etiology): 64% Sensitivity (two most probable etiologies): 81%
Gegundez-Fernandez et al., 2017 (Uvemaster) [35]	Interference method with filtering rules Dataset: 88 cases (all uveitis type)	Demographic (age, gender, immunodeficiency, drugs, trauma or eye surgery, and endemic disease) Ophthalmic (anatomy, chronicity, laterality, granulomatous, vasculitis, papillitis, scleritis, and specific ocular involvement) Clinical examination (skin, mucosal, nervous system, articular, urinary, ear, nose, throat, and digestive and cardiovascular exam) Treatment (steroid response)	88	Sensitivity (most probable etiology): 74% Sensitivity (three most probable etiologies): 91%
Mutawa et al., 2019 [33]	Rule-based expert system (multilayer rule design) Dataset: 61 cases (case report, all uveitis type)	Ophthalmic (anatomy, chronicity, severity, laterality, granulomatous) Treatment (response to therapy)	53	Sensitivity (most probable etiology): 60% Sensitivity (four most probable etiologies): 100%
Jamilloux et al., 2021 [11]	Bayesian belief networks Dataset: 877 cases (all uveitis type)	Demographic (Age, sex, and ethnicity) Ophthalmic (Anatomy, laterality, chronicity, vasculitis, granulomatous, and ocular hypertension)	8	Sensitivity (most probable etiology): 54% Sensitivity (two most probable etiologies): 85%