Table 2 Classification results with the binary classifier of Pain2D for each rare disease versus CP, listing values for true positives (TP), false positives (FP), true negatives (TN), false negatives (FN), p value (Fisher’s exact test), accuracy (Acc), sensitivity (Sens), specificity (Spec), AUC of the ROC curve (AUCROC)
From: A diagnostic support system based on pain drawings: binary and k-disease classification of EDS, GBS, FSHD, PROMM, and a control group with Pain2D
 | TP | FP | FN | TN | p value | Acc | Sens | Spec | AUCROC |
---|
EDS versus CP | 58 | 35 | 1 | 15 | < 0.001 | 0.67 | 0.98 | 0.30 | 0.899 (CI 0.892–0.954) |
GBS versus CP | 28 | 17 | 1 | 33 | < 0.001 | 0.77 | 0.96 | 0.66 | 0.921 (CI 0.853–0.973) |
FSHD versus CP | 34 | 32 | 1 | 18 | < 0.001 | 0.61 | 0.97 | 0.36 | 0.854 (CI 0.77–0.93) |
PROMM versus CP | 80 | 24 | 9 | 26 | < 0.001 | 0.76 | 0.90 | 0.52 | 0.846 (CI 0.774–0.908) |
- In all cases, the p values (Fisher’s exact test) suggest strongly that the classifier performs much better than random guessing