Multivariate learning framework for long‐term adaptation in the artificial pancreas

Figures

• 

  Figure 1. Dual‐layer control scheme. The lower layer (highlighted in blue color) deals with disturbances occurring on smaller time scales based on real time CGM data, while the upper layer (highlighted in green color) handles chronic changes in the glucose metabolic process based on recent performance metrics from the lower layer

• 

  Figure 2. Adaptation of feedforward control parameters

• 

  Figure 3. Trends of the glycemic management metrics in the adaptation procedures for the 111‐patient cohort (Scenario I). A box‐and‐whisker approach is used to plot the data, where on each box, the central white line is the median, the edges of a box denote the 25% and 75% percentiles, and the whiskers denote the 5% and 95% percentiles

• 

  Figure 4. Trends of the adaptation parameters in the adaptation procedures for the 111‐patient cohort (Scenario I). The same box‐and‐whisker approach as that in Figure 3 is used to plot the data. For illustration purpose, the relative values of parameters against default values in the UVA/Padova simulator are provided

• 

  Figure 5. Trends of the glycemic management metrics in the adaptation procedures for the 111‐patient cohort (Scenario II). The same box‐and‐whisker approach as that in Figure 3 is used to plot the data

• 

  Figure 6. Trends of the adaptation parameters in the adaptation procedures for the 111‐patient cohort (Scenario II). The same box‐and‐whisker approach as that in Figure 3 is used to plot the data. Keys are the same as those in Figure 4

• 

  Figure 7. Trends of the glycemic management metrics in the adaptation procedures for the 111‐patient cohort (Scenario III). The same box‐and‐whisker approach as that in Figure 3 is used to plot the data

• 

  Figure 8. Trends of the adaptation parameters in the adaptation procedures for the 111‐patient cohort (Scenario III). The same box‐and‐whisker approach as that in Figure 3 is used to plot the data. Keys are the same as those in Figure 4

• 

  Figure 9. Twenty‐four‐hour glucose and insulin profiles simulated using the system parameters obtained on Weeks 1, 8, 16, and 24 of Scenario I for a particular patient. For comparison purpose, the same meal protocol and measurement noise sequence are applied to generate the data in the four subplots. Green and purple triangles denote meals of 50 g and 75 g CHO, respectively

• 

  Figure 10. Twenty‐four‐hour glucose and insulin profiles simulated using the system parameters obtained on Weeks 1, 8, 16, and 24 of Scenario II for a particular patient. Meal protocol and sensor noise sequence are the same as those in Figure 9

• 

  Figure 11. Twenty‐four‐hour glucose and insulin profiles simulated using the system parameters obtained on Weeks 1, 8, 16, and 24 of Scenario III for a particular patient. Meal protocol and sensor noise sequence are the same as those in Figure 9