Page 22 - 《南京医科大学学报》自然科学版2026年第2期
P. 22
第46卷第2期
·178 · 南 京 医 科 大 学 学 报 2026年2月
在66例患者中进行时间分层验证集,结果也保 校准曲线 Brier 值为 0.122。此外,各数据集中 DCA
持了良好的预测准确性,模型AUC为0.846(95%CI: 均表明该模型在较大阈值概率范围内具有良好的
0.740~0.951),最大约登指数对应的临界值为0.288, 净获益(图4)。
Points
0 10 20 30 40 50 60 70 80 90 100
LVDd(mm)
35 55 75 90
LVEF(%)
70 60 50 40 30
Albumin(g/L)
44 42 40 38 36 34 32 30
Age(years)
60 62 64 66 68 70 72 74 76 78 80 82
Total points
80 100 120 140 160 180 200 220 240 260
Predicted value
0.04 0.08 0.12 0.20 0.40 0.60 0.80
图3 心脏术后衰弱风险的列线图模型
Figure 3 Nomogram for predicting the risk of postoperative frailty after cardiac surgery
A D G
1.00 Cut⁃off=0.247 1.00 Cut⁃off=0.315 1.00 Cut⁃off=0.288
0.75 0.75 0.75
Sensitivity 0.50 Sensitivity 0.50 Sensitivity 0.50
0.25 0.25 0.25 AUC=0.846
AUC=0.846 AUC=0.821
95%CI:0.763-0.928 95%CI:0.701-0.940 95%CI:0.740-0.951
0 0 0
0 0.25 0.50 0.75 1.00 0 0.25 0.50 0.75 1.00 0 0.25 0.50 0.75 1.00
1-Specificity 1-Specificity 1-Specificity
B E H
1.0 Dxy 0.691 Slope 1.000 1.0 Dxy 0.642 Slope 1.000 1.0 Dxy 0.692 Slope 1.000
C(ROC) 0.848 Emax 0.045
C(ROC) 0.846 Emax 0.060
C(ROC) 0.821 Emax 0.121
Actual probability 0.6 Brier -0.020 S:z -0.257 Actual probability 0.6 Brier -0.023 S:z ldeal 0.775 Actual probability 0.6 Brier -0.023 S:z ldeal 0.842
R2
0.037
0.400 E90
0.324 E90
0.094
0.040
R2
0.401 E90
R2
0.045
0.016
0.8 D
0.232 Eavg
0.328 Eavg
0.8 D
0.8 D
0.300 Eavg
0.020
U
U
-0.199
U
-0.286
0.255 S:p
Q
0.322 S:p
0.348 S:p
Q
Q
0.797
0.122
0.144
0.131
Intercept 0.000
Intercept 0.000
Intercept 0.000
0.4
0.4
0.4
ldeal
0.2
Nonparametric
Nonparametric
Nonparametric
0 Logistic calibration 0.2 0 Logistic calibration 0.2 0 Logistic calibration
0 0.2 0.4 0.6 0.8 1.0 0 0.2 0.4 0.6 0.8 1.0 0 0.2 0.4 0.6 0.8 1.0
Predicted probability Predicted probability Predicted probability
C F I
0.4 0.4 0.4
Model Model Model
Net benefit 0.2 0 None Net benefit 0.2 0 None Net benefit 0.2 0 None
All
All
All
0 0.25 0.50 0.75 1.00 0 0.25 0.50 0.75 1.00 0 0.25 0.50 0.75 1.00
Risk threshold Risk threshold Risk threshold
A-C:ROC curve(A),calibration curve(B),and DCA(C)curve for the training set. D-F:ROC curve(D),calibration curve(E),and DCA(F)
curve for the internal validation set. G-I:ROC curve(G),calibration curve(H),and DCA(I)curve for the temporal validation set.
图4 列线图验证
Figure 4 Nomogram validation

