Table 11.
Performance results obtained with 3 classes (crackle vs. wheeze vs. other)—training: fixed duration; testing: variable duration.
| Classifiers | Accuracy | F1Wheez | MCCWheez | F1Crack | MCCCrack | F1Other | MCCOther |
|---|---|---|---|---|---|---|---|
| SUK (Baseline) | 63.3 | 68.1 | 63.5 | 76.8 | 47.1 | 21.7 | 14.6 |
| LDA_10MRMR | 60.3 ± 0.1 | 45.0 ± 0.1 | 42.4 ± 0.0 | 75.1 ± 0.1 | 43.2 ± 0.2 | 36.3 ± 0.0 | 9.8 ± 0.1 |
| LDA_100MRMR | 61.1 ± 0.0 | 69.2 ± 0.5 | 65.0 ± 0.6 | 73.5 ± 0.0 | 39.7 ± 0.0 | 28.8 ± 0.4 | 9.3 ± 0.1 |
| LDA_Full | 62.9 ± 0.3 | 66.6 ± 0.4 | 61.7 ± 0.6 | 76.1 ± 0.4 | 45.4 ± 1.0 | 28.6 ± 1.4 | 13.9 ± 0.6 |
| SVMrbf_10MRMR | 61.9 ± 0.1 | 60.3 ± 2.5 | 54.9 ± 2.3 | 75.7 ± 0.2 | 44.2 ± 0.5 | 31.5 ± 1.2 | 12.0 ± 0.3 |
| SVMrbf_100MRMR | 63.7 ± 0.2 | 68.1 ± 0.4 | 63.3 ± 0.5 | 76.5 ± 0.1 | 46.4 ± 0.4 | 29.1 ± 0.6 | 15.7 ± 0.5 |
| SVMrbf_Full | 63.6 ± 0.5 | 66.9 ± 2.4 | 62.0 ± 2.8 | 77.0 ± 0.2 | 47.5 ± 0.5 | 28.6 ± 1.9 | 14.4 ± 0.8 |
| RUSBoost_10MRMR | 62.1 ± 0.5 | 68.6 ± 1.7 | 64.8 ± 2.4 | 75.8 ± 0.1 | 44.4 ± 0.4 | 20.5 ± 1.3 | 10.5 ± 1.6 |
| RUSBoost_100MRMR | 62.7 ± 0.2 | 69.3 ± 0.8 | 65.5 ± 1.3 | 76.2 ± 0.2 | 45.6 ± 0.5 | 20.3 ± 3.2 | 11.7 ± 1.0 |
| RUSBoost_Full | 62.9 ± 0.4 | 70.8 ± 0.7 | 67.2 ± 0.8 | 76.4 ± 0.6 | 46.1 ± 1.4 | 19.0 ± 4.6 | 11.4 ± 1.8 |
| CNN_dualInput | 61.5 ± 0.4 | 73.0 ± 0.9 | 69.6 ± 1.2 | 75.4 ± 0.4 | 43.6 ± 1.2 | 3.4 ± 0.5 | 3.2 ± 0.9 |
| CNN_Spectrogram | 61.7 ± 0.4 | 71.5 ± 0.8 | 68.0 ± 1.2 | 75.6 ± 0.4 | 43.9 ± 1.1 | 7.5 ± 1.2 | 7.7 ± 0.9 |
| CNN_melSpectrogram | 61.6 ± 0.3 | 72.0 ± 0.8 | 68.8 ± 1.0 | 75.9 ± 0.3 | 44.8 ± 0.9 | 4.3 ± 1.0 | 3.9 ± 1.3 |