Table 2 Summary of the GDL model classification performances.
From: Identification of 12 cancer types through genome deep learning
Cancer type | Raw data | Â | Filter | Accuracy (%) | Sensitivity (%) | Specificity (%) | |||||
|---|---|---|---|---|---|---|---|---|---|---|---|
ALL | Train Data | Test Data | |||||||||
Cancer | Health | Cancer | Health | Cancer | Health | Cancer | Health | ||||
BLCA | 425 | 1991 | 417 | 216 | 341 | 165 | 76 | 51 | 98.43 | 98.68 | 98.04 |
BRCA | 1080 | 1991 | 1073 | 586 | 856 | 471 | 217 | 115 | 98.19 | 97.70 | 99.13 |
COAD | 493 | 1991 | 482 | 842 | 385 | 675 | 97 | 167 | 99.24 | 98.97 | 99.40 |
GBM | 498 | 1991 | 478 | 435 | 385 | 345 | 93 | 90 | 97.81 | 96.77 | 98.89 |
KIRC | 376 | 1991 | 372 | 189 | 288 | 161 | 84 | 28 | 100.00 | 100.00 | 100.00 |
LGG | 530 | 1991 | 518 | 491 | 410 | 397 | 108 | 94 | 99.01 | 99.07 | 98.94 |
LUSC | 561 | 1991 | 545 | 166 | 436 | 133 | 109 | 33 | 100.00 | 100.00 | 100.00 |
OV | 610 | 1991 | 600 | 176 | 481 | 140 | 119 | 36 | 100.00 | 100.00 | 100.00 |
PRAD | 503 | 1991 | 494 | 497 | 399 | 394 | 95 | 103 | 97.47 | 95.79 | 99.03 |
SKCM | 472 | 1991 | 434 | 409 | 358 | 316 | 76 | 93 | 98.22 | 97.37 | 98.92 |
THCA | 504 | 1991 | 503 | 241 | 405 | 190 | 98 | 51 | 97.99 | 97.96 | 98.04 |
UCEC | 561 | 1991 | 549 | 1217 | 446 | 967 | 103 | 250 | 98.02 | 98.06 | 98.00 |
TOTAL | 6613 | 1991 | 5733 | 1629 | 4585 | 1304 | 1148 | 325 | 94.70 | 97.30 | 85.54 |
