Explorer Methodology

How the Exercise IOB Explorer makes its decisions

A plain-language walkthrough of the eight factors behind the Exercise IOB Explorer — what it considers, what the evidence says, and where the model has uncertainty.

The Exercise IOB Explorer is a planning tool, not a real-time estimator. It answers a different question from the Activity Explorer: given a session planned in the future, how much carbohydrate is typically needed beforehand, based on the insulin that will still be active when exercise starts?

It integrates IOB decay calculation with weight-based carbohydrate estimates derived from published evidence — scaled by IOB level, exercise type, duration, and time of day.

Proprietary algorithm notice. The GNL algorithm suite — including the IOB banding thresholds, carbohydrate estimate tables, and exercise-type multipliers — is a UK registered trademark of The Glucose Never Lies® (UK00004267795). Exact band thresholds and gram-per-kg values are proprietary. This page describes the reasoning framework and evidence basis only. Healthcare organisations, app developers, or researchers interested in licensing the algorithm via API should contact john@theglucoseneverlies.com.

The Eight Factors

IOB at Exercise Start

Projecting your active insulin forward to the planned session time

High Confidence

What the model does

You enter your bolus history — units and how long ago each was given. The explorer then projects forward to your planned exercise start time and computes how much insulin will still be active by then. This is the key insight: a meal bolus from 90 minutes ago can still carry significant IOB even when it feels like it has worn off.

Evidence basis

Uses the same biexponential decay model as the Activity Explorer — well-validated pharmacokinetics of rapid-acting analogues. Projecting forward in time uses the same decay function as projecting backward. No additional modelling uncertainty is introduced by the forward projection.

Future Development A future version could allow explicit “no bolus yet” entry for people planning a session before any meal, improving the UX for advance planning scenarios.

Carbohydrate Recommendation from IOB

How active insulin level drives the pre-exercise carb guidance

Moderate Confidence

What the model does

The weight-normalised IOB at exercise start is classified into one of four bands. Each band maps to a directional carbohydrate recommendation: no extra carbs at safe IOB levels, progressively more carbs as IOB rises. At very high IOB, the tool also suggests considering postponing the session. Exact gram amounts are determined by the proprietary algorithm.

Evidence basis

Carbohydrate recommendations for exercise in T1D come from the EASD/ISPAD consensus statement (Riddell et al. Lancet Diabetes Endocrinol 2017;5:377–390), Adolfsson et al. ISPAD 2022, and Moser et al. EASD/ISPAD 2020. IOB/kg banding is consistent with the Activity Explorer. Specific gram amounts are consensus-based estimates — individual ISF and carb ratio variation means they are starting points, not fixed doses.

Clinical Note Carbohydrate estimates in the research are adjusted for individual factors including personal insulin sensitivity factor, exercise history, CGM trend, and last meal timing. The tool outputs are a framework for discussion and education — not a protocol to apply without individual clinical consideration.

Exercise Type Modifier

Aerobic vs resistance vs HIIT — different glucose responses

Moderate Confidence

What the model does

Aerobic exercise gets a higher carbohydrate recommendation than resistance or HIIT at the same IOB level. Aerobic activity lowers glucose most reliably; resistance exercise may raise glucose acutely via stress hormones; HIIT has a mixed initial response followed by sustained lowering. The algorithm applies a directional modifier to scale the base recommendation accordingly.

Evidence basis

The direction of effect is well evidenced. Multipliers are calibrated from published mean glucose responses in exercise RCTs (Bally et al. 2016, Yardley et al., Colberg review 2016). Moderate confidence because between-study variation is high and most trials use controlled conditions that differ from real-world exercise.

Duration Modifier

Short, standard, and extended sessions

Moderate Confidence

What the model does

Longer sessions require more carbohydrate. Sessions are banded into short (under 30 minutes), standard (30–60 minutes), and extended (over 60 minutes). Extended sessions receive an elevated carbohydrate recommendation, reflecting the sustained glucose-lowering effect and the post-exercise insulin sensitivity increase that follows longer aerobic work.

Evidence basis

Duration is a key determinant of glucose fall during aerobic exercise — well established across multiple studies. The cut-off bands at 30 and 60 minutes align with Riddell et al. 2017 guidance and are widely used in clinical exercise protocols for T1D.

Lived Experience Session length in the real world is rarely fixed. People often extend or cut short depending on how they feel or what their CGM shows. The recommendation is a snapshot for a planned duration — re-assessment partway through a longer session is often needed.

Time of Day

Morning insulin resistance and post-meal timing effects

Moderate Confidence

What the model does

Morning exercise (before approximately 10am) is associated with higher hepatic glucose output and elevated cortisol — meaning glucose is less likely to fall and may even rise. The tool applies a directional morning modifier that slightly reduces the pre-exercise carbohydrate recommendation to account for the endogenous glucose contribution from the dawn phenomenon.

Evidence basis

The dawn phenomenon (elevated hepatic glucose output from approximately 4–8am, driven by cortisol and growth hormone) is well established. Its impact on exercise glucose response and circadian carbohydrate requirements is documented by Scott S, Kempf P, Bally L, Stettler C (Nutrients 2019;11(12):3017) — a comprehensive review of circadian factors in T1D exercise management — and Riddell MC et al. 2017. Moderate confidence because magnitude varies considerably between individuals.

Clinical Note Morning exercise is particularly variable — some people find glucose rises sharply regardless of IOB; others see no difference from afternoon sessions. This is worth exploring explicitly with the individual before relying on the time-of-day modifier.

Body Weight Scaling

Per-kg carbohydrate dosing

High Confidence

What the model does

All carbohydrate recommendations are derived from a per-kg basis and then scaled by your actual body weight. A smaller and a larger person with the same relative IOB burden have different absolute carbohydrate requirements. The final gram recommendation reflects this scaling.

Evidence basis

Per-kg carbohydrate dosing is standard clinical and sports nutrition practice. This is the highest-confidence aspect of the explorer — the approach is methodologically sound regardless of the specific IOB thresholds used. Supported by multiple guidelines and consistent with how carb-to-insulin ratios are structured in clinical care.

Carbohydrate Source

Fast, mixed, or slow — timing the carbs to the session

Moderate Confidence

What the model does

You select whether you prefer fast-acting carbs (glucose tablets, sports gels), mixed sources (fruit, sports drinks), or slow-acting sources (oats, bread). The output adjusts the timing guidance accordingly — fast carbs in the final 15–30 minutes before starting; mixed sources 30–45 minutes before; slow carbs 45–60+ minutes before.

Evidence basis

Glucose absorption rates from different carbohydrate sources are well characterised. The timing of pre-exercise carb consumption relative to session start matters significantly for whether glucose is elevated in time to protect against early-session hypoglycaemia.

Future Development A future version could ask for specific food choices and output precise quantities (e.g. “2 Jelly Babies” or “1 Lucozade Sport”) — similar to how the Hypo & Hyper Explorer handles food equivalents. This would improve practical usability for self-management.

Safety Floor

Minimum pre-exercise carbs when glucose is predicted to be low

High Confidence

What the model does

If projected IOB at exercise start is negligible but starting glucose is predicted to be below 5.0 mmol/L, the tool still recommends a fast carb dose. This safety gate triggers on low predicted glucose alone — independent of IOB. It ensures the model does not return a “no carbs needed” output when the glucose trajectory itself poses a risk.

Evidence basis

The 5.0 mmol/L pre-exercise threshold is from Riddell et al. Lancet Diabetes Endocrinol 2017 (EASD/ISPAD consensus). The minimum carb dose aligns with standard hypo treatment guidelines (ADA, Diabetes UK). This combination is the highest-confidence output in the entire explorer — both threshold and dose are directly guideline-consistent.

Clinical Note The safety floor does not account for how accurately the user can predict their glucose at a future point in time. CGM predictions beyond 30 minutes carry inherent uncertainty. If the user is near the 5.0 threshold when planning, a precautionary carb dose should be considered regardless of the model output.

References

Riddell MC, Gallen IW, Smart CE, et al. Exercise management in type 1 diabetes: a consensus statement. Lancet Diabetes Endocrinol. 2017;5(5):377–390.
Adolfsson P, Taplin CE, Zaharieva DP, et al. ISPAD Clinical Practice Consensus Guidelines 2022: Exercise in children and adolescents with diabetes. Pediatr Diabetes. 2022;23(8):1341–1372.
Moser O, Zaharieva DP, Adolfsson P, et al. Glucose management for exercise using CGM and isCGM systems in type 1 diabetes: position statement of EASD and ISPAD. Pediatr Diabetes. 2020;21(8):1375–1393.
Scott S, Kempf P, Bally L, Stettler C. Carbohydrate intake in the context of exercise in people with type 1 diabetes. Nutrients. 2019;11(12):3017.
Yardley JE, et al. Vigorous intensity exercise for glycemic control in patients with type 1 diabetes. Can J Diabetes. 2013;37(6):427–432.
Colberg SR, et al. Physical activity/exercise and diabetes: a position statement of the American Diabetes Association. Diabetes Care. 2016;39(11):2065–2079.
American Diabetes Association. Standards of Medical Care in Diabetes. Diabetes Care. 2024;47 (Suppl 1).
Heise T, et al. Pharmacokinetic and pharmacodynamic properties of insulin analogues. Diabetes Obes Metab. 2015;17(1):1–13.

Disclaimer

This is an educational explorer built from clinical trial data and real-world patterns. It models how algorithms and physiological principles behave on average — not how any individual system will behave for you. It is not a prescription, not a medical device, and must not be used as one. All outputs are for education and discussion only. Any changes to your insulin settings, device configuration, or diabetes management must be made with your diabetes care team.