Table 3.
The depicted confusion matrix illustrates the performance of our proposed diagnostic algorithm to translate measured CD8+ T-cell densities into a respective immune diagnostic category in the validation cohort.
| Confusion matrix | Pathologist diagnosis | True positives (TP) | True negatives (TN) | False positives (FP) | False negatives (FN) | Sensitivity (Recall) | Specificity | Positive predictive value (Precision) | ||
|---|---|---|---|---|---|---|---|---|---|---|
| Desert (n = 10) | Excluded (n = 18) | Inflamed (n = 5) | ||||||||
| Predicted | ||||||||||
| Desert | 10 | 0 | 0 | 10 | 23 | 0 | 0 | 100% | 100% | 100% |
| Excluded | 0 | 15 | 0 | 15 | 15 | 0 | 3 | 83% | 100% | 100% |
| Inflamed | 0 | 3 | 5 | 5 | 25 | 3 | 0 | 100% | 89% | 62.5% |
Whereas the prediction, sensitivity, and specificity of immune desert tumors was very robust, few excluded tumors were attributed to the inflamed immune diagnostic category.