Table 1 Fundus fluorescein angiography dataset characteristics
From: FFA-GPT: an automated pipeline for fundus fluorescein angiography interpretation and question-answer
| Â | Total | Train | Validation | Test | P value |
|---|---|---|---|---|---|
Population | Â | Â | Â | Â | Â |
No. | 9392 | 6312 | 1052 | 2028 | Â |
Age, median (IQR) | 51 (36, 62) | 51 (37, 62) | 51 (36, 61) | 50 (33, 62) | 0.029 |
Sex, n (%) | Â | Â | Â | Â | 0.115 |
Female | 4202 (44.7) | 2795 (44.3) | 502 (47.7) | 905 (44.6) | Â |
Male | 5190 (55.3) | 3517 (55.7) | 550 (52.3) | 1123 (55.4) | Â |
FFA images | Â | Â | Â | Â | Â |
No. | 654,343 | 421,916 | 76,900 | 155,527 | Â |
Phasea, n (%) | Â | Â | Â | Â | <0.001 |
Arterial | 39,844 (6.1) | 27,738 (6.6) | 4794 (6.2) | 7312 (4.7) | Â |
Venous | 372,725 (57.0) | 239,211 (56.7) | 43,778 (56.9) | 89,736 (57.7) | Â |
Late | 241,774 (36.9) | 154,967 (36.7) | 28,328 (36.8) | 58,479 (37.6) | Â |
