/Robo-Endo

Primary LanguagePython

SeniorProject

Short guide to diabetes statistics

Insulin injections last for ~4 hours before being fully absorbed.

Some foods are processed in <1 hours while others take 8+. Diabetics can accound for this by settings the insulin to dispense immediately or overtime. A graph or similar showcasing glucose averages following injections would be useful. We can also target time periods of the injections to figure out if the injections should be sooner, or more delayed relative to food.

IOB should be correlated to carb intake and CGM values to see if there is an abnormal amount of insulin being used. Any rise in IOB without carbs or high CGM is indicative of a hardware failure. Sometimes the CGM will stay high in spite of continuously high IOB, which is also indicative of hardware failure.

Extremely quick CGM changes and erratic are indicative of a CGM failure. This is often characterized by sporatic readings with large gaps and quick direction changes. A 'good' cgm graph has no hard corners.

CGM 'compression failures' are characterized by an increasingly quickly falling CGM reading, normally preceded by no insulin injections. This type of failure usually ends in a gap of time > 10 minutes where the CGM does not read. When the compression failure is fixed (by not pressing your CGM into something), the CGM will pick back up at the previously normal level. Normal flat level > sudden fall to extremely low BS levels(<50) > CGM error, no readings > Normal operation

Exercise and sleep can also affect it. Morning highs are common, especially after breakfast. As well as exercise lows during running or similar.