Different Models For Calculating Insulin On Board

What devices are actually calculating

Devices define IOB using three inputs: the Active Insulin Time (AIT/DIA) setting, a mathematical decay curve, and the total insulin delivered within the time window. None of these inputs know about your body weight, your current insulin sensitivity, or whether the insulin you gave was for food or for a correction. That matters enormously — and it is at the root of almost every frustrating IOB experience.

Carbohydrate absorption: the other moving variable

Devices are trying to match variable insulin with variable glucose appearance. Carbohydrate absorption is not a fixed curve. Simple carbohydrates and high glycaemic foods can lead to rapid glucose appearance, while meals containing fibre, fat, or protein often slow digestion and produce a more gradual rise. Some meals, particularly those high in fat, can delay glucose appearance for several hours. Gastric emptying speed, gut hormones, individual digestive physiology, and conditions such as gastroparesis all further influence the picture.

In other words, we are trying to match variable insulin with variable glucose appearance using a simplified model. Is it any wonder managing Type 1 diabetes can sometimes feel like a proper graft?

The meal insulin vs correction insulin problem

Most systems do not distinguish between insulin given for carbohydrate and insulin given for correction. They are lumped together as IOB. This is problematic because meal insulin is often still covering carbohydrate that has not yet finished digesting. Only the correction insulin is targeting excess glucose in circulation. Physiologically, these behave differently. Mathematically, most devices treat them the same.

High IOB blocks a correction

You take 14 units total: 10 units for food and 4 units for correction. Your AIT/DIA is set to 4 hours and the bolus calculator treats everything as one pool of IOB. Two hours later, when glucose is still entering circulation because your meal has not been fully digested, your glucose is 13.9 mmol/L (250 mg/dL) and rising. You expect a correction. The device says no.

It sees approximately 4 units of total IOB and blocks insulin — even though only approximately 1 unit may be true correction insulin, and the rest is still covering food that has not been absorbed yet. When carbohydrate has been underestimated, which happens 30 to 50% of the time, the algorithm assumes it is preventing stacking by treating all bolus insulin as one pool.

The calculator does not know why insulin was given. It only knows how much was delivered and how much remains within the model. Most of the insulin still circulating is actually meal insulin working against carbohydrate absorption — not free correction insulin, which is the one that matters in this situation.

So what happens in real life? You override. You ghost-carb. You inject by pen. Glucose comes down. You avoid hours of hyperglycaemia. Later, someone says, “Just trust the technology.”

You think: Foxtrot Oscar.

But this is not about trust. It is about modelling. The algorithm sees total risk. Humans think dynamically: food versus correction versus activity. These are the two jobs in conflict.

Why shortening AIT works for tighter glucose management

Consider someone who consistently runs high at 2 to 3 hours post-meal due to carbohydrate underestimation. With AIT set at 5 hours, the system may show substantial IOB and suppress a correction, because it assumes insulin is still active. But if much of that insulin is still dealing with unabsorbed carbohydrate, a correction may in fact be physiologically appropriate. So they shorten AIT to 2 hours, which reduces displayed IOB sooner and allows more aggressive corrections.

On paper, that looks logical. But physiologically, there is still correction insulin working in the background — about 2 units. The result: you may overshoot by roughly 2 units.

Now imagine there is no food involved, just a very high glucose — say 20 mmol/L (360 mg/dL) — and you are correcting every two hours. Because the algorithm treats bolus insulin as effectively gone at 2 hours, it keeps offering full corrections each time. But in reality, some insulin from the previous dose is still active. This is where insulin stacking risk rises quickly.

The predictable results are: unexpected exercise hypoglycaemia, late crashes after stacked corrections, and loss of trust in the bolus calculator. This is not about good or bad settings. It is about what you are optimising for.

Not all IOB models are identical

There are three main approaches to how devices handle IOB in correction calculations.

Model 1: No IOB deduction from corrections

Some systems display IOB but do not use it to reduce correction recommendations. The full calculated correction is always offered. This approach gives maximum correction freedom but carries the highest insulin stacking risk. The displayed IOB, often calculated using a 4 to 6 hour duration, still provides useful information about total circulating insulin for exercise planning — but this information is advisory only, not used by the calculator.

Model 2: Correction-only IOB

Only insulin given explicitly as a correction dose blocks further corrections. Meal boluses are excluded from the correction calculation. This improves correction logic and may reduce stacking from aggressive correction calculations. However, it risks over-correction early after meals, can underestimate true circulating insulin, and may increase hypoglycaemia risk with high-glycaemic meals or when pre-bolusing is inconsistent. It solves one problem but creates others.

Some bolus calculators that exclude meal insulin from correction logic still display both meal and correction insulin on the screen. This transparency helps mitigate the risk. It is essential to check how your system presents this information.

Model 3: Total IOB (meal plus correction pooled)

All insulin — both meal and correction — counts against further corrections. This provides the most stacking protection and the most realistic representation of total circulating insulin for exercise risk. The trade-off is that it can block needed corrections when carbohydrates are underestimated, and may produce persistent hyperglycaemia if corrections are restricted for too long.

The trade-off nobody explains

You can optimise AIT for correction performance (shorter AIT, 2 to 3 hours) or for physiological accuracy (longer AIT, 4 to 6 hours). But you cannot perfectly optimise for both at the same time if meal and correction insulin are grouped together. This is the structural limitation of current IOB models. It is not user error. It is a model architecture decision.

Linear vs curvilinear decay models

Most commercial systems use one of two mathematical decay shapes. Linear degradation means insulin declines evenly over time. Curvilinear degradation means insulin falls more steeply early on, with a longer tail — which more closely mirrors actual pharmacodynamics.

In the curvilinear model, insulin action is front-loaded. The greatest glucose-lowering effect occurs in the first two hours after dosing. After this early peak phase, the rate of effect progressively slows. However, the chosen AIT/DIA duration matters more than the exact curve shape.

Advanced models: offset, peak, and meal-rise settings

Some systems move beyond a simple one decay curve model by introducing the idea of the expected glucose rise after eating. Three settings allow personalisation:

• Meal-rise expectation — an estimate of the peak glucose typical after meals for that individual (1.7 to 5.6 mmol/L or 30 to 100 mg/dL).
• Offset time — at what point does glucose start to decrease from the peak meal-rise (45 to 120 minutes).
• Acting time — an estimate of how long the insulin lasts (3 to 8 hours), creating the drop-off from the end of the offset time.

With this model, corrections during the expected glucose rise phase are restricted, but once the glucose trajectory is expected to start falling, corrections become available again. This means the system avoids blocking corrections when they are genuinely needed — which is a meaningful improvement over pooled IOB models for post-meal management.

AID and hybrid closed-loop systems: an extra layer of complexity

Automated Insulin Delivery systems add a second layer: a user-facing bolus calculator and an internal control algorithm. These layers may not use the same insulin accounting. In practice, displayed IOB may not reflect total insulin exposure generated by automation.

Most systems distinguish between user-entered correction boluses and algorithm-driven micro-corrections, and may apply different decay assumptions to each. When insulin delivery falls below the expected level, this may display as negative IOB — indicating that circulating insulin exposure is lower than anticipated, which can help refine understanding of hypoglycaemia risk.

The key implication: what is displayed on screen may not fully represent what the algorithm is accounting for in the background. Despite these structural differences, the central tension remains. Insulin modelling must balance protection against over-correction while reflecting the front-loaded physiological action of rapid-acting insulin.

The detailed summary table

This table brings together the main IOB modelling approaches used in pumps, bolus calculators, and AID systems, and shows how different configuration choices influence day-to-day behaviour. A simplified version appears at the start of Part 3.