Determination of Aortic Characteristic Impedance and Total Arterial Compliance From Regional Pulse Wave Velocities Using Machine Learning: An in-silico Study

Figures

• 

  FIGURE 1. Schematic representation of the training/testing pipeline for predicting aortic characteristic impedance (Zao), and total arterial compliance (CT). SBP, systolic blood pressure; DBP, diastolic blood pressure; HR, heart rate; cfPWV, carotid-femoral pulse wave velocity; crPWV, carotid-radial pulse wave velocity.

• 

  FIGURE 2. The 1-D cardiovascular model (Reymond et al., 2009) that was used for the data generation.

• 

  FIGURE 3. Typical representation of a random forest regressor.

• 

  FIGURE 4. Typical representation of an artificial neural network.

• 

  FIGURE 5. Learning curves presenting the impact of the number of data instances on the RFR’s performance for Zao (A) and CT (B).

• 

  FIGURE 6. Comparison between predicted and reference data. Scatterplots and Bland–Altman plots between: the predicted Zao and the reference Zao using the RFR (A) and the ANN (C), respectively, and the predicted CT and the reference CT using the RFR (B) and the ANN (D), respectively. The solid line of the scatterplots represents equality. In Bland–Altman plots, limits of agreement (LoA) are defined by the two horizontal dashed lines.

• 

  FIGURE 7. Sensitivity of precision in terms of NRMSE to the number of the training data. The 100% of the training size corresponds to 2,290 cases. NRMSE: normalized root-mean square error.