Table 2. Classification accuracy of RF machine learning algorithm trained on topology indices of parenclictic networks, along with the RF performance on the original gene average methylation level data for different types of cancer.
| Classification with topology indices | Classification with gene methylation | |||||
|---|---|---|---|---|---|---|
| Cancer | Accuracy | Specificity | Sensitivity | Accuracy | Specificity | Sensitivity |
| BLCA | 95.89% | 77.77% | 99.19% | 95.95% | 66.66% | 98.38% |
| BRCA | 96.96% | 91.87% | 98.11% | 97.82% | 87.67% | 98.96% |
| COAD | 99.30% | 94.73% | 100.00% | 99.33% | 94.73% | 100.00% |
| HNSC | 96.70% | 85.57% | 98.69% | 98.60% | 92.27% | 99.36% |
| KIRC | 98.63% | 98.75% | 98.92% | 98.90% | 98.62% | 98.14% |
| KIRP | 99.17% | 97.72% | 100.00% | 96.03% | 95.89% | 95.80% |
| LUAD | 99.43% | 93.75% | 99.01% | 99.39% | 93.72% | 100.00% |
| PRAD | 90.40% | 71.58% | 89.65% | 92.58% | 85.12% | 94.76% |
| THCA | 93.12% | 70.03% | 96.60% | 94.33% | 71.90% | 97.40% |
| UCEC | 98.62% | 91.67% | 99.10% | 99.20% | 91.67% | 100.00% |