Table 1.
Data characteristics of the development set, tuning set, internal validation set and the two external validation sets
| EU development set | EU development set | EU tuning set | EU tuning set | EU internal validation set | EU internal validation set | US external validation set | EU external validation set | Total | |
|---|---|---|---|---|---|---|---|---|---|
| Source | Radboud University Medical Center Netherlands | Karolinska Institutet Sweden | Radboud University Medical Center Netherlands | Karolinska Institutet Sweden | Radboud University Medical Center Netherlands | Karolinska Institutet Sweden | Medical Laboratories, CA/UT, USA; tertiary teaching hospital, CA, USA | Karolinska University Hospital Sweden | – |
| No. of sites | 1 | 1 | 1 | 1 | 1 | 1 | 3 | 1 | 6 |
| No. of cases | 1,028 | 1,085 | 72 | 33 | 129 | 82 | 741 | 330 | 3,500 |
| No. of biopsies | 5,160 | 5,456 | 195 | 198 | 333 | 212 | 741 | 330 | 12,625 |
| Nontumor | 967 (19) | 1,925 (35) | 95 (49) | 58 (29) | 155 (47) | 66 (31) | 254 (34) | 108 (33) | 3,628 (29) |
| Tumor-containing (ISUP GG breakdown below) | 4,193 (81) | 3,531 (65) | 100 (51) | 140 (71) | 178 (53) | 146 (69) | 487 (66) | 222 (67) | 8,997 (71) |
| GG 1 | 852 (17) | 1,814 (33) | 24 (12) | 48 (24) | 48 (14) | 53 (25) | 247 (33) | 65 (20) | 3,151 (25) |
| GG 2 | 675 (13) | 668 (12) | 15 (8) | 32 (16) | 35 (11) | 34 (16) | 122 (16) | 63 (19) | 1,644 (13) |
| GG 3 | 925 (18) | 317 (6) | 15 (8) | 14 (7) | 38 (11) | 16 (8) | 70 (9) | 49 (15) | 1,444 (11) |
| GG 4 | 768 (15) | 481 (9) | 19 (10) | 30 (15) | 16 (5) | 22 (10) | 21 (3) | 19 (6) | 1,376 (11) |
| GG 5 | 973 (19) | 251 (5) | 27 (14) | 16 (8) | 41 (12) | 21 (10) | 27 (4) | 26 (8) | 1,382 (11) |
| No. of cases with general pathologist reviews | – | – | – | – | 70 | – | 237 | – | 307 |
| No. of pathologist reviews | – | – | – | – | 910 | – | 4,740 | – | 5,650 |
The values in parentheses give the percentage. The development set was available to competition teams for algorithm development, and the tuning set for limited algorithm evaluation during the competition. All validation sets were fully independent and blinded to the algorithm developers. Additional details on reference standard protocol can be found in Supplementary Methods 2 and 3.