Advances in Non-Invasive Blood Pressure Monitoring

Figures

• 

  Figure 1. BoppliTM device to monitor the blood pressure of infants in critical care.

• 

  Figure 2. Schematic of operating principle of PyrAmes’ capacitive sensor technology.

• 

  Figure 3. Comparison of normalized invasive arterial line data (red) taken simultaneously with normalized PyrAmes sensor data (blue).

• 

  Figure 4. Proof-of-concept results for a single person convolutional neural network algorithm trained on cuff data collected from March through May and then tested in September and October.

• 

  Figure 5. Results from convolutional neural network model trained with cuff data from Study 1 and tested with an ambulatory blood pressure monitor for 4 other ambulatory subjects (SBP values color coded by individual).

• 

  Figure 6. Plots of convolutional neural network model outputs vs. invasive arterial line ground truth values of mean arterial pressure (MAP), systolic (SBP), and diastolic (DBP) blood pressure values averaged over each individual. The points are color coded for the patients’ age in days.

• 

  Figure 7. Bland-Altman plots of BP-model-3 results for mean arterial (MAP), systolic (SBP), and diastolic (DBP) blood pressure values in mmHg averaged over each individual. The x-axis is the average of the model and reference blood pressure values. The y-axis is the difference between the model and reference values. The points are color coded for the patients’ weight in kilograms. The red solid line indicates the overall error of the model (Mean Difference (MD)). The dotted red lines indicate 1.96 × the standard deviation (sd). The green dotted lines indicate the MD limits for accuracy of ±5 mmHg per guidelines from the US Food and Drug Administration.

• 

  Figure 8. Comparison of invasive arterial line (blue) and BP-model-3 (red) mean arterial pressure (MAP) values as a function of time. Each data point is averaged over a one-minute interval. The curves are a spline fit with lambda = 1 × 10−6.