Nomogram Based on HRV for Predicting the Therapeutic Effects of Orthostatic Training in Children with Vasovagal Syncope

Figures

• 

  Figure 1. Workflow of the research performed to estimate the therapeutic effect of orthostatic training in children with vasovagal syncope.

• 

  Figure 2. Factors chosen in the lasso regression for estimating the therapeutic effect of orthostatic training in children with VVS. (a) Lasso coefficient profile plot. The regression coefficient of each independent variables varied as λ changed. The color lines stand for different variables. As λ increased, the model exhibited a higher compression degree. (b) Cross-validation plot for the penalty term. The vertical line on the left corresponds to the λ value that yields the minimum mean squared error (MSE), while the vertical line on the right corresponds to the λ value that yields one standard error away from the minimum MSE. The optimal value was determined by selecting the left vertical line for the investigation.

• 

  Figure 3. The visual predictive model developed using a nomogram for estimating the therapeutic effect of orthostatic training in children with VVS. rMSSD, the root means square of successive differences between normal sinus beats. For example, for a 13-year-old patient with an rMSSD of 26 ms, the corresponding points (red dots) were acquired by drawing vertical lines through the points representing the values of the two parameters and perpendicular to the axes of the points (red dashed line), resulting in approximately 40 points. The corresponding probability is 0.316 (red double arrow); therefore, orthostatic training is not recommended.

• 

  Figure 4. Model evaluation of the training dataset for estimating the therapeutic effect of orthostatic training in children with VVS. (a) Receiver operating characteristic (ROC) curves. (b) Calibration curve. The x and y axes represent the predicted probabilities and observed probabilities. The line labeled “ideal” is the reference line indicating perfect calibration, whereas the line labeled “apparent” represents the performance of visual predictive model. The intercept is 0.00 (95% CI, −0.54–0.54), slope is 1.00 (95% CI, 0.49–1.51), and c-statistic is 0.81 (95% CI, 0.69–0.89). The bias-corrected line represents the visual predictive model performance corrected by bootstrapping (B = 1000 repetitions). (c) Decision curve analysis (DCA). The y-axis indicates the net benefit, and the x-axis indicates the threshold probability. The line labeled ‘none’ corresponds to the scenario where no patients receive orthostatic training, whereas the line labeled ‘all’ corresponds to the scenario where all patients receive orthostatic training.

• 

  Figure 5. Model evaluation of the validation dataset for estimating the therapeutic effect of orthostatic training in children with VVS. (a) Receiver operating characteristic (ROC) curves. (b) Calibration curve. The x and y axes represent the predicted probabilities and observed probabilities. The line labeled “ideal” is the reference line indicating perfect calibration, whereas the line labeled “apparent” represents the performance of visual predictive model. The intercept is −0.12 (95% CI, −0.82–0.58), slope is 0.89 (95% CI, 0.32–1.45), and c-statistic is 0.80 (95% CI, 0.63–0.90). The bias-corrected line represents the visual predictive model performance corrected by bootstrapping (B = 1000 repetitions). (c) Decision curve analysis (DCA). The y-axis indicates the net benefit, and the x-axis indicates the threshold probability. The line labeled ‘none’ corresponds to the scenario where no patients receive orthostatic training, whereas the line labeled ‘all’ corresponds to the scenario where all patients receive orthostatic training.