bg_ISF determination and its impact

The evidence suggests a correlation between higher glucose levels and the need for a stronger ISF* . Successful users of AAPS have established automation rules that enhance the ISF profile at elevated glucose levels. However, a drawback is observed with abrupt changes in ISF at switch points, lacking further adaptations in between.

In the context of autoISF, a polygon defines the relationship between glucose and ISF, interpolating values in between. Although currently hard-coded, users can apply weights to effortlessly adjust its strength, aligning with personal requirements. The polygon itself is editable, allowing users to rebuild the .apk (AAPS) for a different shape if necessary. While contemplating the development of a GUI for this purpose, the effort was deemed tedious, with uncertainty about the results justifying such an undertaking. This approach enables an approximation of the formula used in DynamicISF for the ISF dependency on glucose.

Two weighting factors come into play based on whether glucose is below or above the target:

lower_ISFrange_weight: Used below the target, it weakens ISF the more the weight increases. A weight of 0 disables this contribution, maintaining a constant ISF below the target. This weight is less critical, especially when the loop is likely running at TBR=0. A starting point could be around 0.2.

higher_ISFrange_weight: Used above the target, it strengthens ISF the more the weight increases. A weight of 0 disables this contribution, keeping ISF constant above the target. Start with a weight of 0.2, monitor reactions, and check the SMB-tab before cautiously increasing it. The result is expressed as follows:

bg_ISF = 1 + xxx_ISFrange_weight * glucose_polygon_Lookup

A special case exists when bg_ISF < 1, indicating a condition below the target. In this scenario, ISF is weakened, and further checks for remaining effects are unnecessary. Positive acceleration mitigates the weakening, signaling an impending rise in glucose levels.