Table 3.
Performance of AI Systems for the Detection of DR with Portable Retinal Cameras (Selected Studies)5,5, 23, 24, 25
| Authors and Year | Population/Setting | Sample Size (Rate of DR) | Output | Sensitivity/Specificity (%) | Camera, AI System |
|---|---|---|---|---|---|
| Lupidi et al,5 2023 | Specialized Retina Clinic (Italy) | 256 (50%) | Any DR | 96.8/96.8 | Optomed Aurora (Optomed, Oulu, Finland); Selena+ |
| Ruan et al,23 2022 | Hospital-based (China) | 315 (N/A) | Referable DR | 88.2/40.7 | Optomed Aurora (Optomed, Oulu, Finland); Phoebus |
| Rajalakshmi et al,24 2018 | Hospital-based (India) | 296 (65%) | Any DR | 95.8/80.2 | Remidio (Remidio Innovative Solutions, India); EyeArt |
| Sight-threatening DR | 99.1/80.4 | ||||
| Nunez do Rio et al,25 2022 | Community screening, rural and urban areas (India) | 11 199 (3.8%∗) | Referable DR | 72.08/85.65 | Zeiss Visuscout (Carl Zeiss Meditec, Jena, Germany); VISUHEALTH-AI DR |
AI = artificial intelligence; DR = diabetic retinopathy; N/A = not available.
∗
Referable DR.