Table 2.
Random forest model in predicting lung cancer against controls and other cancers.
| CTL vs LC | Benign vs LC | BC vs LC | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Predicted | Predicted | Predicted | |||||||
| Group | CTL | LC | Group | Benign | LC | Group | BC | LC | |
| Actual | CTL | 9 | 1 | Benign | 28 | 12 | BC | 10 | 7 |
| LC | 1 | 9 | LC | 1 | 32 | LC | 2 | 31 | |
| Error | 0.1 | Error | 0.178 | Error | 0.18 | ||||
| Sensitivity | 90% | Sensitivity | 96.97% | Sensitivity | 93.94% | ||||
| Specificity | 90% | Specificity | 70% | Specificity | 58.82% | ||||
| CCA vs LC | CRC vs LC | EC vs LC | |||||||
| Predicted | Predicted | Predicted | |||||||
| Group | CCA | LC | Group | CRC | LC | Group | EC | LC | |
| Actual | CCA | 18 | 7 | CRC | 12 | 10 | EC | 12 | 2 |
| LC | 1 | 32 | LC | 1 | 32 | LC | 1 | 32 | |
| Error | 0.138 | Error | 0.2 | Error | 0.064 | ||||
| Sensitivity | 96.97% | Sensitivity | 96.97% | Sensitivity | 96.97% | ||||
| Specificity | 72% | Specificity | 54.55% | Specificity | 85.71% | ||||
| GC vs LC | |||||||||
| Predicted | |||||||||
| Group | GC | LC | |||||||
| Actual | GC | 38 | 9 | ||||||
| LC | 1 | 32 | |||||||
| Error | 0.125 | ||||||||
| Sensitivity | 96.97% | ||||||||
| Specificity | 80.85% | ||||||||
Sensitivity = number of true positives / (number of true positives + number of false negatives); Specificity = number of true negatives / (number of true negatives + number of false positives); CTL, healthy controls; LC, lung cancer; BC, bladder cancer; CCA, cervical cancer; CRC, colorectal cancer; EC, esophageal cancer; GC, gastric cancer. All test sets except for the test set 1 (CTL vs LC) compares all lung cancer patients to benign diseases or other cancers.