Page 458 - Handbook of Biomechatronics
P. 458

452                                                    Graham Brooker




                     Closed-loop
            Real-time  control      State update   Data filtering  User interface
             control  algorithm                                (meal info etc.)
             module
                 Recommended
                   basal rate
                   differential
                      Safety
            Continuous                                         User interface
             safety  supervision   State estimation  Data filtering  (bolus approval etc.)
             module   algorithm
                                                     Delivered insulin  CGM data  Basal rate
                   Approved
                      total
                     insulin
                                    Data logging
             Interface  Actuation   and outcome    Data handling  User interface
             module                  measures                 (patient info etc.)




                                                      Continuous
                      Insulin
                       pump                            glucose
                                                       monitor
          Fig. 30 The MPC prediction scheme relies on past inputs and outputs with future
          outputs predicted as a function of future inputs within the control horizon. (Based on
          Cobelli, C., DallaMan, C., Sparacino, G., Magni, L., DeNicolao, G., Kovatchev, B., 2009.
          Diabetes: models, signals and control. IEEE Rev. Biomed. Eng. 2: 54–96.)




          which improvements in accuracy have little effect on clinical outcomes. To
          date only one system CGM system meets this criterion (Castle et al., 2017).
             There is a drive toward producing CGM systems that are accurate
          enough without finger-prick calibrations for both control and insulin shut
          off. Commercial systems developed by Dexcom and Medtronic obtained
          FDA approval for both of these functions in 2016. Novel CGM systems
          include implantable types discussed earlier, and flash glucose monitoring
          (CGM on demand). The latter can replace finger-prick measurements
          completely in some cases.
             One of the issues surrounding the incorporation of sophisticated control,
          safety, and multihormonal delivery into the insulin pump is that it compli-
          cates what would otherwise be a comparatively simple device into a complex
          power-hungry computer with a pump attachment. An alternative is to keep
          both the CGM and the pump as simple as possible and provide wireless com-
          munication to a smart phone as shown in Fig. 31. This approach leverages a
          powerful, yet cheap computing platform with performance driven by a mas-
          sive user base beyond diabetics. This allows for new and improved
   453   454   455   456   457   458   459   460   461   462   463