v0.0.9

Performance optimizations and converting tests be more fixture-based

v0.0.8

Performance improvements

Version 0.0.7

• Add a command to calculate a schedule of unabsorbed carbohydrates, also known as carbs on board (COB).
• Improve accuracy of IOB for temp basals of irregular duration (e.g 3.2 minutes)

v0.0.6

• Adding --start-at and --end-at limits for IOB output
• Fixing numeric parsing for openaps report args

v0.0.5

• Adding a new command, glucose_momentum_effect for applying a linear regression to recent glucose history
• Adding --momentum parameter to glucose_from_effects to blend momentum effect into the prediction
• Fixing an issue where scheiner_carb_effect absorption parameters could not be read from an openaps report serialization
• Adding support for a legacy Nightscout API format (#31)

Note that scipy is now a dependency. It's recommended you use your favorite package manager to install rather than building from source.

v0.0.4

• Separates the single glucose use into its individual components: walsh_insulin_effect, scheiner_carb_effect, and glucose_from_effects. The glucose command remains for convenience and compatibility, but its implementation simply runs through the other components.
• Adds the walsh_iob command to calculate total insulin on-board from normalized dosing history.
• Adds support for timezone-aware timestamps. It is up to the user to ensure that either all pump & glucose history is timezone-aware or that it is all timezone-naïve. Mixing the two will raise an exception as comparisons become unclear.

v0.0.3

Adds support for "event" units, now present in openapscontrib.mmhistorytools==0.1.0

v0.0.2

• Fixes an incorrect application of a sensor effect delay
• Adds some safety assertions that history was recently written and glucose data is current. NOTE: This will require you to ensure your system timezone is the same as your glucose sensor time.