Page 68 - 南京医科大学自然版
P. 68

第44卷第8期
               ·1096 ·                           南 京    医 科 大 学 学         报                        2024年8月


              患者显著肝纤维化的决策具有临床获益(图4)。模                           维化的风险。
              型的净获益曲线(红色)位于无干预(None,蓝色横                         2.4.3 aCHANGE模型与FIB⁃4模型预测能力的比较
              线)和全面干预(All,绿色斜线)的极端情况之上,说                             在训练集以及验证集中计算出 FIB⁃4,计算公
              明aCHANGE模型比这两种极端假设更能为临床提                          式:FIB⁃4=(AST×年龄)/(血小板计数× ALT ),后
              供价值,有助于准确预测NAFLD患者出现显著肝纤                          利用 FIB⁃4 对患者是否出现显著肝纤维化进行预


                       A                                          B
                           1.0  Dxy  0.550                             1.0  Dxy  0.551
                              C(ROC)  0.775                                C(ROC)  0.775
                              R2    0.239                                  R2   0.235
                              D U Q  -0.001                                D U Q  0.153
                                    0.158
                          Actual probability 0.8  Brier  0.122        Actual probability 0.8  Brier  -0.246
                                                                                0.001
                                                                                0.153
                                    0.157
                                                                                0.123
                                    0.000
                                                                           Intercept
                              Intercept
                                                                       0.6
                           0.6
                                                                                0.881
                              Slope
                                                                           Slope
                                    1.000
                                    0.045
                                                                           Emax
                              Emax
                                                                                0.147
                                                                           E90
                                    0.014
                                                                                0.054
                              E90
                                                                           Eavg
                                                                                0.015
                                    0.007
                              Eavg
                                                                       0.4
                           0.4
                                                                                0.222
                                    0.193
                                                                           S:z
                              S:z
                           0.2  S:p  0.847   Ideal                     0.2  S:p  0.824   Ideal
                                             Logistie eaboration                         Logistie eaboration
                                             Nonparametric                               Nonparametric
                            0                                           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
                                  A:Calibration curve of the training set. B:Calibration curve of the internal validation set.
                                                 图2   aCHANGE模型的校准曲线
                                          Figure 2 Calibration curve of the aCHANGE model
                       A                                          B
                           1.0                                         1.0
                           0.8                                         0.8
                          Sensitivity  0.6                            Sensitivity  0.6
                                                                       0.4
                           0.4
                           0.2                                         0.2
                                              AUC:0.775                                   AUC:0.775
                            0                                           0
                              .0   0.2  0.4  0.6  0.8  1.0                .0   0.2  0.4  0.6   0.8  1.0
                                       1-Specificity                                1-Specificity
                 A:ROC curve of the training set(area under the curve approximately 0.775). B:ROC curve of the internal validation set(area under the curve ap⁃
              proximately 0.775).
                                                 图3 aCHANGE模型的ROC曲线
                                            Figure 3 ROC curve of the aCHANGE model
                       A                                          B
                          0.2    Model_step                            0.2  Model_step
                                 All                                        All
                                 None                                       None
                          Net benefit  0.1                            Net benefit  0.1
                           0                                            0
                             00    0.25   0.50  0.75   1.00                    0.25    0.50   0.75
                                      Risk threshold                               Risk threshold
                                    A:Clinical DCA on the training set. B:Clinical DCA on the internal validation set.
                                             图4   aCHANGE模型的临床决策曲线分析
                                           Figure 4  Clinical DCA of the aCHANGE model
   63   64   65   66   67   68   69   70   71   72   73