Preview - two sides of A4 landscape. Ctrl/Cmd+P → A4 Landscape · Margins None · Background graphics ON → save as PDF.
Survive & Thrive · Exercise Series

Exercise on AID Systems : what each algorithm does, what you still do

Side 1 of 2
System-specific settings
All UK AID Systems
Exercise framework

AID handles some of the work. You still do the rest.

IOB
IOB is still variable #1 Every AID algorithm works by modulating basal delivery - but they don't cancel insulin already delivered. Pre-meal bolus reductions still apply.
Pre-exercise lead
60-90 min
activate target / Ease-off
Bolus reduction
25-50%
meal within 90 min of exercise
Post-exercise
up to 4-14 h
late drop risk window
Ref: Moser/Zaharieva 2024 (EASD/ISPAD AID consensus) · Royston & Hovorka 2024 CamAPS · Campbell 2015 · Rabasa-Lhoret 2001.
The three majors (AID context)
✓1 · IOB. AID can reduce future basal but cannot reclaim delivered insulin. Pre-meal bolus matters.
✓2 · Starting glucose. Same as MDI and pump - where you start shapes where you land.
✓3 · Trend arrows. Arrows tell you what the algorithm is seeing. Read them, don't wait.
✓Activate exercise mode before the session. 60-90 min of lead time lets IOB drop to working range.
✓Don't turn it off during. Algorithm behaviour is designed for full session coverage.
AID system · exercise mode reference

What each system does, what you still do

SystemExercise modeWhat changes
Omnipod 5Activity ModeTarget → 8.3 mmol/L, basal reduced
Control-IQExercise ModeTarget 7.8-8.9 mmol/L, no auto-correct
MiniMed 780GTemp TargetTarget → 8.3 mmol/L, less basal, no auto-correct
CamAPS FXEase-offReduces insulin delivery broadly
All four do similar things. Raise target, reduce basal, disable strong auto-corrections. The timing advice is the same: activate before, not during.
Pre-meal bolus · still yours

What the algorithm doesn't cover

  • Meal within 90 min of exercise: reduce bolus 25-50% (moderate session), 50-75% (longer / harder).
  • Exercise > 90 min after meal: bolus is largely cleared - switch to carb intake during exercise.
  • AID will auto-correct a rise - but post-meal excursions that are too big for the algorithm push you out of working range.
Pre-exercise starting glucose · AID-adjusted

The target has shifted - use it

Glucose at startAction (exercise mode already on)
< 5.0 mmol/L15-30 g carb, delay starting 15 min.
5.0-7.010-15 g carb. Algorithm target is above this - it will reduce basal.
7.0-10.0Start as planned. Inside most exercise-mode targets.
10.0-15.0Start. Algorithm will auto-correct if target is exceeded.
> 15.0Check ketones. If present, delay. If not, light session only.
CamAPS Boost + Ease-off · real-world data

A note on Boost, not just Ease-off

Royston & Hovorka 2024 (n=7,464 CamAPS users): Boost (for illness, growth, persistent highs, pregnancy) had 0.0% time below 3.9 mmol/L vs 2.1% outside Boost. Ease-off reduced hyperglycaemia during exercise.

Both are safe across all ages. Boost does not cause hypoglycaemia. Ease-off does not cause hyperglycaemia. Use them.
Survive & Thrive · Exercise Series

Exercise on AID Systems : during, after, and overnight

Side 2 of 2
Carbs, recovery & overnight
All UK AID Systems
During exercise

Carbs + what the algorithm is doing

AID is already reducing insulin delivery - but algorithm response lags physiology. Carbs still bridge the gap.

  • Light 30-60 min: 5-15 g every 30 min if glucose drifts down.
  • Moderate 30-60 min: 15-20 g every 30 min.
  • Heavy / prolonged > 60 min: 30-60 g/hour, higher if aerobic.
  • Intermittent high-intensity: glucose often rises from sprints. AID will auto-correct - don't fight it with carbs.
The algorithm can't see carbs you eat. For mid-session carbs, don't bolus them if they're there to prevent hypo. Do enter them if they're a planned fuel source during extended sessions.
Non-carb counter-regulation

10-second sprint + AID

Bussau 2006: 10-sec maximal sprint at the end of moderate exercise attenuates post-exercise glucose drop. On AID the effect is additive - the algorithm responds to the sprint-induced rise by briefly increasing delivery, smoothing what would otherwise be a spike.

Iscoe & Riddell show the same for intermittent high-intensity intervals within continuous moderate exercise - AID handles the variability well.

Post-exercise & overnight

The late drop is still real - algorithms help, don't fix

Post-exercise glucose dip arrives 4-14 hours later. AID algorithms reduce delivery during the dip - but don't always catch a fast fall quickly enough.

  • Keep exercise mode on 60 min post-session. Let the algorithm reduce basal while IOB still high.
  • Bedtime snack 0.4 g/kg slow carbs (Campbell 2015) even on AID. Don't bolus if glucose is in range.
  • Mixed-macronutrient snack (milk + oat biscuit) outperforms quick carbs for overnight stability (Kalergis 2003).
  • CamAPS Ease-off can run overnight to reduce overall delivery. Review with your team.
  • First few sessions of a new activity: set a 3am alarm until you know your pattern, even on AID.
Major in the majors · the two actions

Model before you act

Action 1 Exercise Planning Explorer - AID-awarePick your AID system - explorer models exercise-mode timing + bolus + carbs for that algorithm. → Action 2 Exercise IOB - carbs for 30 min exerciseIOB on AID still matters. Calculates what the algorithm cannot claw back. →
Go deeper

GNL explorers, guides & podcasts

GuideFull Exercise Guide - 5-part seriesIncluding Part 4 - AID + exercise, Part 5 - Mastering Top 10.→ GuideFull AID Guide - 5-part seriesHow every algorithm handles exercise differently.→ AskGrace - evidence-backed Q&AMoser/Zaharieva 2024, Royston & Hovorka CamAPS, pregnancy.→