A Novel Ultrafiltration Rate Feedback Controller for Use in Hemodialysis: First Clinical Experience

Figures

• 

  Figure 1. Proportion of RBV values above, within and below the respective RBV target range for each of the RBV target time points during study treatments with the Controller. Underlying data: all subjects who contributed data (n=14), all RBV targets (300 data points). RBV, relative blood volume.

• 

  Figure 2. Proportion of treatments with RBV within the desired target range. Underlying data: subjects who contributed at least two complete treatments to each study period and whose contributing standard-of-care treatments were within 90 days of the study period (n=9 subjects). Open circles denote individual per-subject data, filled circles denote mean across all subjects, and error bars denote SEM. UFR, ultrafiltration rate.

• 

  Figure 3. Density plot of intradialytic SBP, stratified by study period (standard-of-care versus UFR Feedback Controller). Underlying data: all intradialytic SBP values from subjects who contributed BP measurements to both study periods and whose contributing standard-of-care treatments were within 90 days of the study period (n=14 subjects). Both complete and incomplete treatments are included. SBP, systolic BP.

• 

  Figure 4. Proportion of Controller suggestions disregarded by the health care staff for all 48 complete study visits. The percentage of Controller suggestions disregarded is denoted above each bar. The two numbers along the bottom indicate how many suggestions were disregarded out of how many suggestions in total. The numbers along the top of the chart denote the number of suggestions that were disregarded for acute medical reasons, defined as low BP, clinical symptoms, or both.

• 

  Figure 5. Distribution of the differences between the Controller-suggested UF goals (top panel) and rates (bottom panel) and those implemented by the health care staff. Underlying data: the 25 study visits (from 11 subjects) where the staff chose to disregard Controller suggestions. In the bottom panel, the reasons for disregarding the Controller suggestions are shown for the four largest differences between the suggested and implemented UF rates. UF, ultrafiltration.

• 

  Figure 6. Example study treatment (subject UFC-010, visit 1). The solid black line represents the subject's filtered RBV curve (see Methods for details). Raw RBV values are shown in light gray. The red line denotes the desired RBV trajectory targeted by the Controller. The green boxes indicate the half-hourly RBV ranges associated with significantly improved patient survival, as published by Preciado et al.8 The dashed black line shows the UF rates suggested by the Controller. The plot title further provides the absolute AUC (pp×minutes) and average absolute difference between the filtered RBV curve and the desired RBV trajectory. AUC, area under the curve; pp, percentage points.