Skip to main content
. 2020 Dec 24;21(1):57. doi: 10.3390/s21010057

Table 14.

Performance results obtained with 3 classes (crackle vs. wheeze vs. other)—training: variable duration; testing: variable duration.

Classifiers Accuracy F1Wheez MCCWheez F1Crack MCCCrack F1Other MCCOther
LDA_10MRMR 62.3 ± 0.1 71.0 ± 0.0 67.8 ± 0.0 75.2 ± 0.1 42.5 ± 0.1 17.1 ± 0.2 14.2 ± 0.3
LDA_100MRMR 65.5 ± 0.0 72.3 ± 0.1 69.8 ± 0.1 76.7 ± 0.1 47.8 ± 0.1 35.0 ± 0.4 22.5 ± 0.1
LDA_Full 68.8 ± 0.1 72.2 ± 0.1 69.9 ± 0.1 78.2 ± 0.2 52.9 ± 0.2 48.9 ± 0.5 32.5 ± 0.3
SVMrbf_10MRMR 65.6 ± 0.4 72.5 ± 0.5 69.1 ± 0.7 76.7 ± 0.3 47.2 ± 0.7 34.9 ± 2.7 23.2 ± 1.3
SVMrbf_100MRMR 68.2 ± 0.9 68.8 ± 1.9 64.1 ± 2.2 77.4 ± 0.7 51.2 ± 1.4 52.1 ± 2.3 31.3 ± 2.2
SVMrbf_Full 68.0 ± 1.1 65.2 ± 4.0 60.9 ± 3.9 75.9 ± 1.6 51.1 ± 1.5 57.7 ± 3.2 33.4 ± 1.7
RUSBoost_10MRMR 65.4 ± 0.4 72.7 ± 0.6 69.9 ± 0.8 74.8 ± 0.9 45.2 ± 0.8 43.2 ± 3.8 24.1 ± 1.6
RUSBoost_100MRMR 68.5 ± 0.5 73.6 ± 0.8 71.0 ± 1.2 75.4 ± 1.3 50.6 ± 1.0 55.2 ± 2.5 33.6 ± 1.1
RUSBoost_Full 69.0 ± 1.1 73.7 ± 0.7 70.7 ± 0.7 75.4 ± 1.6 51.6 ± 1.7 57.7 ± 0.6 35.2 ± 1.6
CNN_dualInput 81.8 ± 0.7 72.5 ± 2.3 69.3 ± 2.0 88.2 ± 0.6 75.2 ± 1.3 75.1 ± 1.3 62.1 ± 1.4
CNN_Spectrogram 78.7 ± 0.9 70.5 ± 3.0 66.3 ± 3.1 86.2 ± 0.6 70.9 ± 1.7 69.6 ± 2.6 55.9 ± 1.8
CNN_melSpectrogram 76.9 ± 1.3 70.3 ± 2.6 66.2 ± 2.4 84.7 ± 0.8 67.4 ± 2.0 66.3 ± 3.9 51.4 ± 3.0