Table 3 Validation performance of the top-performing two-gene decision trees that were trained using the discovery dataset (Mathys 2019)
Removed | Selected | Accuracy | Specificity | Sensitivity | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Mathys 2019 | Zhou 2020 | Xiong 2023 | Mathys 2024 | Mathys 2019 | Zhou 2020 | Xiong 2023 | Mathys 2024 | Mathys 2019 | Zhou 2020 | Xiong 2023 | Mathys 2024 | ||
- | UQCRHL, ETS2 | 99% | 94% | 99% | 98% | 99% | 99% | 99% | 99% | 84% | 3% | 18% | 45% |
ETS2 | UQCRHL, CDKN2D | 99% | 94% | 99% | 97% | 99% | 99% | 99% | 99% | 83% | 3% | 16% | 46% |
UQCRHL | CDKN2D, ETS2 | 99% | 98% | 99% | 99% | 100% | 100% | 100% | 99% | 83% | 77% | 63% | 89% |
UQCRHL, ETS2 | CDKN2D, RB1 | 99% | 97% | 99% | 97% | 100% | 100% | 99% | 97% | 81% | 62% | 72% | 92% |
UQCRHL, ETS2, CDKN2D | E2F3, RB1 | 99% | 96% | 98% | 94% | 100% | 100% | 98% | 94% | 67% | 43% | 67% | 95% |
UQCRHL, ETS2, RB1 | CDKN2D, CDK6 | 99% | 96% | 99% | 98% | 100% | 100% | 99% | 98% | 77% | 49% | 56% | 90% |
UQCRHL, ETS2, RB1, CDK6 | CDKN2D, ATM | 99% | 97% | 99% | 97% | 100% | 100% | 99% | 97% | 70% | 65% | 66% | 91% |
UQCRHL, ETS2, RB1, CDK6, CDKN2D | E2F3, RBL2 | 98% | 96% | 99% | 96% | 100% | 100% | 99% | 96% | 58% | 41% | 54% | 93% |
UQCRHL, ETS2, RB1, CDK6, ATM | CDKN2D, RBL2 | 99% | 97% | 99% | 98% | 100% | 100% | 99% | 98% | 73% | 65% | 58% | 90% |
UQCRHL, ETS2, RB1, CDK6, ATM, RBL2 | CDKN2D, ATP6V1D | 99% | 97% | 99% | 98% | 99% | 99% | 99% | 98% | 88% | 68% | 51% | 88% |
UQCRHL, ETS2, RB1, CDK6, ATM, RBL2, ATP6V1D | CDKN2D, E2F3 | 99% | 96% | 99% | 98% | 100% | 100% | 100% | 99% | 66% | 44% | 54% | 88% |
UQCRHL, ETS2, RB1, CDK6, ATM, RBL2, ATP6V1D, E2F3 | CDKN2D, CES4A | 99% | 97% | 99% | 98% | 99% | 99% | 99% | 98% | 82% | 69% | 37% | 84% |
