Table 4.
Experimental results on clinical datasets (%, average (std) of cross-validation)
| Method | Accuracy | AUC | F1 | Recall | Precision | p-value |
|---|---|---|---|---|---|---|
| Ori CNN | 84.80(2.43) | 89.00(1.65) | 70.29(4.26) | 63.46(3.51) | 78.83(5.70) | <0.05 |
| MC-CNN | 84.51(1.29) | 90.85(1.13) | 70.55(1.29) | 62.85(1.53) | 80.84(4.42) | <0.05 |
| LSTM | 86.27(1.29) | 90.27(1.15) | 74.17(2.47) | 69.73(2.62) | 79.56(5.69) | 0.08 |
| Time-LSTM | 85.79(2.37) | 90.81(1.57) | 74.57(3.81) | 71.08(3.56) | 78.71(6.48) | 0.42 |
| tLSTM | 86.42(1.48) | 91.06(1.48) | 74.36(1.99) | 68.55(1.55) | 81.49(5.28) | 0.40 |
| DLSTM1 | 86.97(1.45) | 91.17(1.53) | 76.11(2.68) | 72.71(2.38) | 80.04(5.18) | *(base) |
| DLSTM2 | 86.98(1.20) | 91.41(1.51) | 75.54(1.67) | 71.24(5.01) | 81.22(6.11) | -- |
| DLSTM3 | 85.99(1.13) | 91.10(1.69) | 74.68(2.89) | 70.51(6.07) | 80.23(5.55) | -- |
| DLSTM4 | 86.91(1.37) | 91.07(1.28) | 75.85(1.94) | 72.39(3.65) | 80.21(6.34) | -- |
*
The average and standard deviation (std) of five-fold test results are reported.
*
The best average results are shown in bold. The p < 0.05 indicates our method significantly improve the compared method (McNemar test).
*
Methods citations: Ori CNN (Liao et al., 2019), LSTM(Hochreiter and Schmidhuber, 1997) (Shi et al., 2015), Time-LSTM (Zhu et al., 2017), tLSTM(Santeramo et al., 2018).