Table 7.
Confusion matrix and per-class performance achieved by the proposed method across all five-folds for the atrial fibrillation (AFIB) classification task on the 2017 PhysioNet/CinC Challenge database.
Predicted |
Per-class Performance (%) |
||||||||
---|---|---|---|---|---|---|---|---|---|
N | A | O | ~ | acc | ppv | sen | spec | ||
Actual | N | 4221 | 53 | 738 | 63 | 78.65 | 81.83 | 83.17 | 71.98 |
A | 70 | 463 | 207 | 18 | 93.75 | 66.05 | 61.08 | 96.93 | |
O | 839 | 172 | 1348 | 53 | 75.83 | 57.51 | 55.89 | 83.70 | |
~ | 57 | 13 | 51 | 157 | 97.01 | 53.95 | 56.47 | 98.37 |
acc: accuracy; ppv: positive predictive value; sen: sensitivity; spec: specificity