Table 5 Results of the proposed method compared to generic classifiers on 365 patients with various etiologies [21, 22, 24, 25]
From: Justified granulation aided noninvasive liver fibrosis classification system
Method | Overall accuracy [%] | Specificity [%] | Sensitivity [%] | ||||
|---|---|---|---|---|---|---|---|
Ŝ 1 | Ŝ 2 | Ŝ 3 | Ŝ 1 | Ŝ 2 | Ŝ 3 | ||
Granular model | 67.4 | 84.2 | 91.3 | 91.4 | 84.2 | 38.5 | 78.3 |
Random Tree | 63.4 | 74.8 | 76.5 | 92.1 | 69.1 | 54.0 | 67.0 |
PNN | 61.6 | 75.7 | 90.9 | 75.7 | 67.1 | 39.3 | 81.4 |
Naïve Bayes | 60.8 | 60.2 | 84.7 | 92.9 | 80.1 | 38.5 | 59.7 |
Auto-Binned IBk | 58.4 | 59.8 | 79.4 | 94.7 | 73.9 | 40.1 | 57.7 |
RecBF-DDA | 61.4 | 77.6 | 83.1 | 92.5 | 63.6 | 48.3 | 74.2 |
IBk (k-NN) | 44.8 | 31.5 | 76.5 | 100.0 | 81.5 | 31.9 | 0.0 |