skip to Main Content

July 25, 2025

Leads:

Carol Levy – Mt. Sinai (Adult)

Emily Coppedge – Cornell (Peds)

Dr. Rayhan Lal, MD (Stanford Medicine) Discussed DIY Systems

  • Discussion on key takeaways, research findings, and implications for clinical practice
  • Q&A with Dr. Rayhan Lal

Open-Source Automated Insulin Dosing

  • Progress in the field has been slow.
  • The twist system is a commercial open-source algorithm.
  • The first widely available automated insulin dosing system was open APS.
  • Options include Medtronic’s (up to 780G), Control IQ, Eyelet, and Omnipod 5.

Challenges with Automated Insulin Dosing

  • Data from the Omnipod 5 showed that only 28.3% of kids and 44.3% of adults were meeting both glucose targets.
  • Our fastest insulins are not much faster than our last gen insulins.
  • The time to when people bolus is more deviant than the delta that arises from faster insulins.

Regulatory and Data Ownership Barriers

  • The Class 2 regulatory environment at the FDA introduces special controls established by the company that develops the first device in their category.
  • People with diabetes don’t own their data.
  • HIPAA is obsolete because it defines health information as data residing in a designated record set maintained by a covered entity (health providers or plan), not large biotech companies.
  • Manufacturers are typically considered business associates, so device data does not enter the designated record set and exists in a legal gray area.
  • Medtech corporations act as gatekeepers, controlling access for people with chronic diseases, healthcare systems, researchers, and startups.

 Commercial Algorithm Comparison

  • Medtronic: Most advanced algorithm, worst usability.
  • Tandem: Middle of the road usability and algorithms.
  • Omnipod 5: Best usability, worst algorithm.
  • Open source fills the gap of best usability with the best algorithms.

Loop, IAPs, and Trio

  • Compatible pump models: Old Medtronic pods, Omnipod Eros, Omnipod Dash.
  • CGM options are available.
  • Trio and Loop share pump and CGM drivers.
  • Libre: XSTrip for iOS to listen in on LibreLink.
  • Dexcom: Corresponding Dexcom apps.

Loop Algorithm

  • The loop algorithm is straightforward and uses similar math to predict future glucose levels.
  • Future predicted glucose = IOB * ISF downward + (carbs on board / carb ratio) * ISF.
  • ISF over carb ratio is also known as carb sensitivity ratio.
  • Carb sensitivity ratio is well preserved for people at different ages.
  • Loop uses short-term adaptation through retrospective correction.
  • If the prediction is wrong, the delta is added to the next prediction.
  • Adaptation occurs over 5-15 minutes.
  • Loop may not be ideal for unannounced meals because adaptation takes time.
  • Highs may occur initially, followed by lows due to delayed pullback.

Therapy Settings

  • Glucose target should be set.
  • Start with one setting that works most of the time.
  • Additional profiles can be added later if needed.
  • Carb ratio: 0.8
  • ISF: 40

Open APS Filter

  • Open APS has a filter to avoid spurious points being taken as unannounced meals.
  • The default filter only accounts for glucose rises of 20%.
  • A glucose rise of 40% is generally okay unless there is a noisy G7.

 Algorithm Settings

  • Enable dynamic settings, specifically dynamic ISF, set to Sigmoid.
  • Weighted T should be 65%, and adjust basal toggle should be on.
  • Dynamic ISF requires a certain number of days of data (up to 7 days).
  • Dynamic ISF is beneficial for unannounced meals.

Adjustments

  • Adjustments can be made through overrides or temp targets.
  • Overrides:
  • Example: “Boost” preset at 110% with target kept as is.
  • Can save commonly used presets like “walking.”
  • Walking preset: Set to 70%, target at 120, and disable micro boluses.
  • Adjust ISF and carb ratio.

Loop

  • Turn on help integration and choose whether to share data.
  • Set Night Scout glucose safety limit (equivalent to 85).
  • Correction range is the target range.
  • Carb ratios and basal rates need to be set.
  • People new to these systems need assistance with physiology and diabetes-specific settings.
  • Assist with supplies and titration.

 Choosing a System: Loop, Trio, or iAPS

  • Loop is better for Type A personalities who never miss a bolus.
  • Trio or APS is better for those who want diabetes to be secondary to the rest of their life.
  • These systems are good for those who don’t always remember to bolus.
  • Android APS and Trio/iAPS support unannounced meals.

Other topics that were covered:

  • Delivery Limits
  • Open-Source Aid
  • Terminology
  • Nightscout (the first tool for sharing CGM data).
  • Trio Setup
  • Open APS Algorithm
  • Free APS and Forks
  • Time in Range
  • Pump Hardware

Recording: Hybrid Closed Loop Working Group-20250725_140446-Meeting Recording.mp4

 

 

T1DX-QI

This Post Has 0 Comments

Leave a Reply