Table 2. Training and Validation Datasets for Diabetic Retinopathy on Training, Primary Validation, and 10 External Validation Datasetsa.
Datasets | No. | No. (%) | ||||||||
---|---|---|---|---|---|---|---|---|---|---|
Nonreferable Eyes | Referable Eyesb | |||||||||
Patients | Images | Eyes | No Diabetic Retinopathy | Mild Nonproliferative Diabetic Retinopathy | Moderate Nonproliferative Diabetic Retinopathy | Severe Nonproliferative Diabetic Retinopathyc | Proliferative Diabetic Retinopathyc | Diabetic Macular Edema | Ungradable | |
Training | ||||||||||
Singapore National Diabetic Retinopathy Screening Program 2010-2013,d | 13 099 | 76 370 | 38 185 | 33 709 (88.3) | 3310 (8.7) | 597 (1.6) | 478 (1.3) | 70 (0.2) | 2026 (5.3) | 21 (0.1) |
Primary Validation | ||||||||||
Singapore National Diabetic Retinopathy Screening Program 2014-2015,e | 14 880 | 71 896 | 35 948 | 33 087 (92.0) | 1808 (5.0) | 455 (1.3) | 170 (0.5) | 24 (0.1) | 320 (0.9) | 404 (1.1) |
External Validation | ||||||||||
Community-based | ||||||||||
Guangdongf | 3970 | 15 798 | 7899 | 5665 (71.7) | 1235 (15.6) | 737 (9.3) | 0 | 154 (1.9) | 0 | 108 (1.4) |
Population-based | ||||||||||
Singapore Malay Eye Study | 763 | 3052 | 1526 | 1143 (74.9) | 215 (14.1) | 113 (7.4) | 18 (1.2) | 9 (0.6) | 53 (3.5) | 28 (1.8) |
Singapore Indian Eye Study | 1128 | 4512 | 2256 | 1639 (72.7) | 422 (18.7) | 125 (5.5) | 5 (0.2) | 17 (0.8) | 71 (3.1) | 48 (2.1) |
Singapore Chinese Eye Study | 484 | 1936 | 968 | 759 (78.4) | 131 (13.5) | 60 (6.2) | 1 (0.1) | 7 (0.7) | 17 (1.8) | 10 (1) |
Beijing Eye Study | 263 | 1052 | 526 | 493 (93.7) | 4 (0.8) | 11 (2.1) | 4 (0.8) | 0 | 12 (2.3) | 2 (0.4) |
African American Eye Disease Study | 492 | 1968 | 984 | 807 (82.0) | 50 (5.1) | 37 (3.8) | 5 (0.5) | 16 (1.6) | 28 (2.85) | 41 (4.17) |
Clinic-based | ||||||||||
Royal Victoria Eye and Ear Hospital | 588 | 2302 | 1151 | 432 (37.5) | 121 (10.5) | 159 (13.8) | 123 (10.7) | 191 (16.6) | 249 (21.6) | 125 (10.9) |
Mexican | 343 | 1172 | 586 | 38 (6.5) | 284 (48.5) | 192 (32.8) | 51 (8.7) | 18 (3.1) | 223 (38.1) | 3 (0.5) |
Chinese University of Hong Kong | 314 | 1254 | 627 | 224 (35.7) | 114 (18.2) | 235 (37.5) | 43 (6.9) | 11 (1.8) | 96 (15.3) | 0 |
University of Hong Kongg | 1932 | 7706 | 3853 | 1984 (51.5) | 1485 (38.5) | 155 (4.0) | 14 (0.4) | 0 | 214 (5.55) | 1 (0.03) |
Total | 38 253 | 189 018 | 94 509 | 79 980 (84.9) | 9179 (9.74) | 2876 (3.04) | 912 (0.97) | 517 (0.55) | 3309 (3.50) | 791 (0.41) |
For study locations and race/ethnicity data, see Table 1.
Referable diabetic retinopathy is defined as moderate nonproliferative diabetic retinopathy, severe nonproliferative diabetic retinopathy, proliferative diabetic retinopathy, and diabetic macular edema.
Vision-threatening diabetic retinopathy is defined as severe nonproliferative diabetic retinopathy and proliferative diabetic retinopathy.
Nine patients had only 1 eye.
In the Singapore Diabetic Retinopathy Screening Program 2014-2015, there were 6291 patients who were repeats from the Singapore Diabetic Retinopathy Screening Program 2010-2013, and 8589 were unique patients.
Forty-one patients had only 1 eye.
Eleven patients had only 1 eye.