Table 1.
Clinical performance of AQR and time-domain ventricular late potentials (VLP) parameters.
| Parameters | Number of Parameters | SP (%) | SE (%) | TPA (%) | AUC (%) |
|---|---|---|---|---|---|
| Time-domain VLP parameters | |||||
| fQRSD | 1 | 64.3 | 73.3 | 68.1 | 71.4 |
| RMS40 | 1 | 64.3 | 90.0 | 75.0 | 86.7 |
| LAS40 | 1 | 71.4 | 70.0 | 70.8 | 81.1 |
| Linear combination of VLP parameters | |||||
| fQRSD + RMS40 + LAS40 | 3 | 73.8 | 73.3 | 73.6 | 83.8 |
| Individual AQR parameters | |||||
| AQR_X(20, 12) | 1 | 81.0 | 60.0 | 72.2 | 75.3 |
| AQR_Y(20, 9) | 1 | 90.5 | 73.3 | 83.3 | 84.6 |
| AQR_Z(20, 8) | 1 | 76.2 | 73.3 | 75.0 | 77.3 |
| AQR_X(26, 10) | 1 | 78.6 | 70.0 | 75.0 | 74.4 |
| AQR_Y(32, 10) | 1 | 85.7 | 80.0 | 83.3 | 81.1 |
| AQR_Z(34, 10) | 1 | 85.7 | 76.7 | 81.9 | 81.3 |
| Linear combination of AQR parameters | |||||
| AQR_X(20, 1:20) | 20 | 85.7 | 80.0 | 83.3 | 88.6 |
| AQR_Y(20, 1:20) | 20 | 92.9 | 80.0 | 87.5 | 93.6 |
| AQR_Z(20, 1:20) | 20 | 88.1 | 80.0 | 84.7 | 93.3 |
| AQR_X(20, 1:20) + AQR_Y(20, 1:20) + AQR_Z(20, 1:20) | 60 | 97.6 | 86.7 | 93.1 | 99.1 |
| AQR_X(2:40, 10) | 20 | 85.7 | 86.7 | 86.1 | 92.5 |
| AQR_Y(2:40, 10) | 20 | 83.3 | 80.0 | 81.9 | 89.4 |
| AQR_Z(2:40, 10) | 20 | 88.1 | 80.0 | 84.7 | 92.4 |
| AQR_X(2:40, 10) + AQR_Y(2:40, 10) + AQR_X(2:40, 10) | 60 | 100 | 90 | 95.8 | 99.4 |
AQR_l (M, σ) denotes the AIQP-to-QRS ratio estimated by an RBFNN with the neuron number M and the spread parameter σ in lead l, AQR_l (20, 1:20) denotes 20 AQR parameters estimated by an RBFNN with σ = 1, 2, …, 20, while keeping M constant at 20, and l represents the X, Y or Z lead, and AQR_l (2:40, 10) denotes another 20 AQR parameters estimated by an RBFNN with M = 2, 4, …, 40, while keeping σ constant at 10. SP, SE, TPA and AUC denote specificity, sensitivity, total prediction accuracy and the area under the receiver operating characteristic curve, respectively.