Sarah Withee
@geekygirlsarah
Diabetes - When that process doesn't work
Type 1 - Auto-immune disease, Insulin-making cells
destroyed
Type 2 - Cells are resistant to use what you do
make
There are other rarer types too
In US, measured in mg/dL (milligrams per deciliter)
Normal people: 70-99 mg/dL
Diabetics aim for: 80-150 mg/dL
After eating, the number goes up
Glucose too high (hyperglycemia)?
Slowly get bad circulation problems,
organs slowly die
Glucose too low (hypoglycemia)?
Shaking, sweating, weakness, could land
in hospital
1 “unit” of insulin = 1/100 mL
Carb ratio: number of carbs / unit of insulin taken
Can be from 3g-50g per unit (3:1 to 50:1)
Varies by person, time of day, conditions, and over lifespan
Carbs start absorbing within
10-20 mins
Insulin absorbs slowly
Fastest around 45-60 mins
Absorbs totally over 5 hours
(blue line)
Stressed?
Sick?
Exercise?
Little bit of alcohol?
Lots of alcohol?
Hormones?
Glucose goes up
Glucose goes up
Glucose goes down
Glucose can go up
Glucose can go down
Glucose can go up/down
Stressed?
Sick?
Exercise?
Little bit of alcohol?
Lots of alcohol?
Hormones?
Glucose goes up
Glucose goes up
Glucose goes down
Glucose can go up
Glucose can go down
Glucose can go up/down
Continuous
glucose
monitor
Insulin pump
OpenAPS - Open Artificial Pancreas System
CGM sends values over Bluetooth to xDrip+
Stored in Nightscout
Values load into OpenAPS rig
Rig sends RF to pump to dose differently
Live Demo (sorta)!
URL: clicky
SSHing into "Sar-borg"
tl;dr It's not a REAL pancreas, but replicates the functions
tl;dr yes
tl;dr anything can be hacked, but it's too obnoxious/slow to do
tl;dr kinda, but worth it
tl;dr analytics stop, but pump goes back to default
Before Sar-borg
After Sar-borg
OpenAPS-based system:
- Instructions: https://openaps.readthedocs.io
- OpenAPS: github.com/openaps/oref0
- Nightscout: github.com/nightscout/cgm-remote-monitor
- xDrip+: github.com/NightscoutFoundation/xDrip
LoopKit-based:
- Instructions: loopkit.github.io/loopdocs/
- LoopKit: github.com/LoopKit/Loop
- RileyLink: github.com/ps2/rileylink
What if our software could change lives?
- Young type 1 children and their parents
- Older people
- Low income
- Those without health insurance