Exercise and Type 1 Diabetes

AID Systems and Exercise: Key Guidance

Automated insulin delivery (AID) systems transform day-to-day glucose management. During exercise, the same automation that works so well at other times can make sessions feel less predictable — because the algorithm has been adjusting insulin in the background for hours before activity even starts. Understanding why this happens is the first step to working with it.

AID Systems Exercise Insulin

This content is for educational exploration only. It describes average responses and general principles. It is not medical advice and cannot replace individual clinical guidance from your diabetes care team.

Overview

This page explores how to manage exercise when using an AID system (closed loop / hybrid closed loop), following the GNL principle of major in the majors. For exercise with AID, the dominant drivers are still insulin on board, then starting glucose, then trend arrows. Device modes and exercise type are important, but they come after.

New international guidance on exercise with AID systems is available: the EASD–ISPAD consensus statement.

Supplementary graphics from the paper:

System-specific one-page reference guides:

The three majors

Major 1: Insulin on board — still the main driver

AID does not remove the core physiology of exercise in type 1 diabetes: exercise amplifies insulin action. Whether using injections, a standard pump, or an AID system, exercise-related glucose drops are typically a problem of too much active insulin for the session being undertaken.

The mechanism here is important: working muscle takes up glucose more readily, and the insulin already circulating becomes more effective. More insulin on board means a larger glucose-lowering effect — and a greater risk of overshooting the lower end of range.

Sessions where insulin on board is low at the start tend to be the most reproducible and easiest to learn from. A pragmatic anchor used in clinical practice is the Three-Hour Rule: where possible, avoid starting planned exercise within approximately three hours of a significant bolus, or consider reducing that bolus if activity is planned or unavoidable.

AID can help reduce insulin on board at the start of a session — but only if given enough time. Increasing glucose targets or enabling activity features 1–2 hours before planned exercise gives the algorithm time to reduce insulin delivery and lower active insulin levels before the session begins.

Major 2: Starting glucose value

Starting glucose acts as a physiological buffer. Starting low-normal leaves less margin for the glucose drops that are common during aerobic activity. Starting high can prompt the algorithm to increase insulin delivery — which can make the drop more pronounced once movement begins.

A common pattern with AID: glucose creeps up pre-session (from stress, warm-up, or pre-exercise carbohydrate), the system responds with more insulin, and then the session begins with more active insulin than intended — leading to a drop larger than expected.

Awareness of this pattern is worth exploring. Many people find that modest pre-exercise carbohydrate, combined with an early target increase, helps avoid this cascade.

Major 3: Trend arrows — direction and speed

During exercise, trend arrows often carry more actionable information than the glucose number itself. A stable 7.0 mmol/L (126 mg/dL) is physiologically different from 7.0 mmol/L trending downward at speed.

A heuristic from clinical practice: supplementing with small amounts of fast-acting carbohydrate — often in the 3–20 g range every 20–30 minutes — is a commonly observed approach when glucose is around 7.0 mmol/L (126 mg/dL) and trending down during exercise. The appropriate amount varies considerably with intensity, duration, and individual insulin on board.

Using glucose value and trend direction together — rather than either alone — is the approach reflected across current clinical guidance for in-session decisions.

Why AID can make exercise feel less predictable

AID systems adjust insulin frequently in the background — that is exactly why they work so well overnight and day-to-day. The downside during exercise is reproducibility.

Two sessions that look identical on paper — same time of day, same activity type, same starting glucose — may have been preceded by very different insulin delivery patterns. The algorithm may have been more or less aggressive in the hours before each session, without the user being aware. That makes trial-and-error learning harder, because “same session” does not mean “same starting physiology”.

When manual open-loop can improve reproducibility

One strategy used in clinical practice, particularly for people who want to develop a reliable personal exercise playbook, is to temporarily step out of full automation to increase repeatability.

How this approach typically works

A manual basal rate matched to the planned session is set in advance.
Switching to open-loop (or a less automated mode) approximately 2 hours before the session allows insulin conditions to stabilise.
That stable basal is maintained into and during the activity, with small carbohydrate adjustments guided by glucose value and trend arrows.
Returning to closed loop after exercise is the common pattern — particularly to regain overnight AID benefits.

This is not a “better” way to live with an AID system — it is a way to make learning more systematic: stable starting conditions lead to clearer pattern recognition, which builds a more reliable personal playbook. Many people then return to full automation once they have that understanding.

Where exercise type fits

Exercise type matters, but it sits below insulin on board, starting glucose, and trend arrows in the hierarchy of factors to consider.

Aerobic exercise tends to lower glucose during and after the session, with a prolonged effect on insulin sensitivity.
High-intensity or sprint work may initially raise glucose (driven by adrenaline and counterregulatory hormones), with a later drop once hormones settle — often 30–60 minutes after the session ends.
Resistance training tends to produce smaller immediate glucose changes but meaningful delayed effects on insulin sensitivity, particularly overnight.

The art with AID is layering device tools — targets, modes, activity features — on top of these physiological fundamentals, rather than trusting the algorithm to “figure it out” without any input.

System-specific features: what they actually do

Each system has different tools for exercise. The goal across all of them is the same: reduce insulin on board at the start of activity, then manage glucose using trend-informed carbohydrate and appropriate targets or modes.

Beta Bionics iLet

Limited ability to adjust glucose targets pre-exercise, so carbohydrate planning and real-time supplementation during activity become particularly important. Discuss a personal exercise protocol with your diabetes team.

CamAPS FX

“Ease-off” mode raises targets and reduces algorithm aggressiveness. Most effective when activated well in advance — typically 1–2 hours before exercise — so insulin on board is lower when the session starts.

MiniMed 780G

“Temp Target” raises the target glucose level and stops automated correction boluses during exercise. This can help prevent repeated glucose lows once movement begins, particularly when there is significant insulin on board from a pre-session bolus.

Omnipod 5

“Activity” mode sets a higher target and makes the algorithm less aggressive. When activated early enough — typically 1–2 hours before the session — it reduces insulin on board before activity begins.

Tandem t:slim X2 with Control-IQ

“Exercise Mode” shifts targets upward and moderates insulin delivery. Most useful when enabled in advance for sessions that tend to lower glucose — aerobic exercise in particular.

DBLG1

“Physical Activity” mode for exercise; “ZEN” mode can raise the protective target for longer or less predictable sessions.

Evidence-based patterns for starting a personal framework

The following reflects patterns observed in clinical practice and published guidance. Individual responses vary considerably — these are population-average starting points for exploration and discussion with a diabetes care team.

For planned exercise: enabling the system’s activity mode or increasing the target 1–2 hours before the session is a common approach for reducing insulin on board at the start — consistent with EASD–ISPAD guidance.
If exercise is within approximately 2 hours of a meal: in clinical practice, reducing the carbohydrate entry into the system by around 25–33% is a commonly used approach to limit bolus insulin given. The algorithm cannot reverse insulin already delivered.
For unplanned exercise: setting a higher target promptly, and consuming small amounts of fast-acting carbohydrate when glucose is around 7.0 mmol/L (126 mg/dL) and trending down, reflects the approach seen in published guidance for this scenario on average.
During exercise: small, repeated carbohydrate doses guided by value and trend arrows tend to produce better outcomes on average than large reactive corrections — this is consistent with ISPAD 2022 and EASD–ISPAD 2025 exercise guidance.
If exercise feels chaotic: a reproducibility block using manual open-loop for planned sessions — stable basal, clearer pattern recognition — is a framework described in clinical practice to support pattern learning before returning to closed loop.
After exercise: delayed glucose effects are well documented, particularly after longer sessions. Insulin sensitivity may remain elevated for 12–24 hours after sustained activity — a pattern highlighted in multiple CGM-based exercise studies.
CGM accuracy at low glucose: if symptoms and sensor reading diverge, a confirmatory finger-prick is the more reliable approach — consistent with all CGM manufacturer guidance.

For one-page summaries, the system-specific guides linked at the top of this page are worth keeping accessible alongside the EASD–ISPAD consensus graphics.