How Does Omnipod 5 Work? Automated Insulin Delivery

The Omnipod 5 is an automated insulin delivery system that continuously adjusts background insulin based on readings from a continuous glucose monitor, effectively creating a feedback loop between glucose levels and insulin delivery without requiring tubing or manual pump adjustments throughout the day. It pairs a small, wearable insulin pod with a Dexcom G6 or G7 sensor and a control algorithm embedded in the pod itself, making real-time dosing decisions every five minutes. The technology sits at the intersection of wearable hardware and adaptive software, and the clinical evidence behind it has grown rapidly since its U.S. launch.

The Three Parts of the System

At its core, the Omnipod 5 relies on three components working in concert. The first is the Pod, a small, adhesive device worn directly on the skin that holds up to 200 units of rapid-acting insulin. Unlike traditional insulin pumps, it has no tubing. A tiny cannula inserts automatically when the Pod is applied, and the Pod delivers insulin subcutaneously from there. You wear each Pod for up to three days before replacing it.

The second component is a continuous glucose monitor, specifically the Dexcom G6 or G7 sensor worn separately on the body. This sensor reads interstitial glucose levels and transmits that data wirelessly to the Pod every five minutes. The CGM is the system’s eyes: without a steady stream of glucose readings, the algorithm cannot make its adjustments.

The third piece is the algorithm itself, which lives on a chip inside the Pod rather than on a separate device. This is a meaningful design choice. Because the algorithm runs on the Pod, it continues making insulin adjustments even if your phone dies or you walk away from it. The smartphone app (or a dedicated controller) is used to enter meals, change settings, and view data, but the Pod keeps working autonomously as long as it has CGM data.

How the Algorithm Makes Decisions

The algorithm uses a model-predictive control approach, which means it does not just react to your current glucose level. It projects where your glucose is heading over the next several hours and adjusts insulin delivery to steer you toward a target range. Every five minutes, the system recalculates how much basal insulin to deliver, increasing it when glucose is rising or predicted to rise and decreasing or suspending delivery when glucose is falling or predicted to drop too low.

You set a glucose target, and the system works to keep you near that number. The available targets range from 110 to 150 mg/dL in increments of 10. A lower target like 110 mg/dL results in more aggressive insulin delivery to keep glucose tighter, while a higher target like 150 mg/dL gives the algorithm a more relaxed posture. Real-world data from nearly 70,000 users showed that people who chose a target of 110 mg/dL spent about 69% of their time in the 70–180 mg/dL range, while those with targets of 130–150 mg/dL spent roughly 54% of time in range. Time spent below 70 mg/dL remained under about 1% across all target settings.¹

The system does not replace bolus insulin for meals. When you eat, you still need to tell the system how many carbohydrates you are consuming so it can deliver a mealtime bolus. What it automates is the background basal delivery, the steady drip of insulin that covers non-meal metabolic needs. If you under-count carbs at a meal and your glucose starts climbing higher than expected, the algorithm will increase basal delivery to help bring it back down, but this correction is slower than a properly sized meal bolus would be.

How It Learns Over Time

One of the more distinctive features of the Omnipod 5 algorithm is that it adapts to your individual insulin needs as you use it. The system tracks your total daily insulin and recalculates a personalized estimate with each Pod change, using a weighted average from the previous four to five Pods.² This means the algorithm becomes more attuned to your patterns over roughly the first two weeks. During the first Pod, the system relies on whatever basal program you or your clinician entered. After that, it starts incorporating actual delivery data.

This adaptive quality has a practical consequence: the first few days on the system are often the roughest. The algorithm is still learning, and users may see more glucose variability than they expect. Clinicians who onboard patients onto the system generally advise patience during this initial period and warn against making dramatic setting changes before giving the algorithm time to calibrate.³

The Activity Feature and Exercise

Exercise is one of the trickiest challenges for anyone on insulin. Physical activity increases glucose uptake by muscles independently of insulin, which means you can drop low quickly if you have too much insulin on board. The Omnipod 5 includes an Activity Feature that you can activate before exercise. When turned on, it raises the algorithm’s effective glucose target, prompting it to reduce insulin delivery in advance of and during physical activity.

A randomized controlled trial tested this feature by comparing standard automated mode to activating the Activity Feature either 30 or 60 minutes before exercise. Both pre-activation windows significantly reduced insulin delivery in the hour before and during exercise. The glucose drop during exercise was also smaller: participants in standard auto mode saw an average glucose drop of about 57 mg/dL during exercise, compared to roughly 44 mg/dL when the feature was activated 60 minutes ahead and about 36 mg/dL when activated 30 minutes ahead.⁴ The rates of hypoglycemia during exercise were lower with the Activity Feature enabled, though the differences did not reach statistical significance in that trial, likely because of the sample size.

The takeaway for users is straightforward: if you plan to exercise, activating the feature at least 30 minutes beforehand gives the algorithm time to dial back insulin. Activating it 60 minutes early works too, though the trial found the 30-minute window was similarly effective at blunting the glucose drop. Waiting until you are already mid-workout is less helpful because rapid-acting insulin that was already delivered is still active in your body.

Clinical Results in Type 1 Diabetes

The evidence base for the Omnipod 5 in type 1 diabetes now spans both controlled trials and large real-world datasets. A study of nearly 70,000 users found that adults using the tightest glucose target of 110 mg/dL achieved a median time in range of about 70%, while children and adolescents on the same target reached around 65%.¹ Those numbers are meaningful because the international consensus target for time in range is 70% or above, and many people with type 1 diabetes struggle to get there with injections or conventional pumps.

A separate study looking at people transitioning onto the system found that time in range improved by about 16 percentage points on average, with HbA1c dropping significantly. The biggest improvements showed up in people who started with higher HbA1c values: those who began above roughly 7.5% saw the largest reductions.⁵ This pattern is consistent across diabetes technology studies generally. The people with the most room for improvement tend to benefit the most, while someone already achieving tight control may see smaller gains.

Expansion Into Type 2 Diabetes

The Omnipod 5 was originally cleared for type 1 diabetes, but its use in type 2 diabetes has been actively studied. The SECURE-T2D trial enrolled adults with type 2 diabetes who were already on insulin, either by injection or pump therapy. After 13 weeks on the Omnipod 5, average HbA1c dropped from 8.2% to 7.4%, and time in the 70–180 mg/dL range increased from 45% to 66%, a gain of roughly 20 percentage points or about 4.8 extra hours per day spent in target range. The improvement was driven primarily by reductions in time spent at high glucose levels. Time spent in hypoglycemia did not increase, and there were no cases of diabetic ketoacidosis during the trial.⁶

This is a notable development because type 2 diabetes affects far more people than type 1, and many of them end up on insulin as the disease progresses. Historically, insulin pump technology and automated systems have been almost exclusively a type 1 domain. The idea that a closed-loop system could also meaningfully help insulin-requiring type 2 patients opens a significant new frontier for the technology.

How It Compares to Other Closed-Loop Systems

The Omnipod 5 is not the only automated insulin delivery system on the market. The Medtronic 780G and the Tandem Control-IQ (t:slim X2 with CIQ) are two other widely used systems. All three connect a pump to a CGM and use an algorithm to adjust basal insulin. The key hardware difference is that the Omnipod 5 is tubeless, while the Medtronic and Tandem systems use traditional tubed pumps with separate infusion sets.

A comparative study of 176 participants using Medtronic’s system, Control-IQ, or Omnipod 5 found that all three achieved similar time-in-range outcomes at 90 days: roughly 67–70% across the groups, with no statistically significant difference. HbA1c reductions were somewhat larger with Control-IQ and Omnipod 5 compared to the Medtronic system, and time spent below range was lowest with Omnipod 5.⁷ The differences between systems were modest. For most people, the practical choice between these systems comes down to factors like whether you prefer tubeless wear, which CGM sensor is compatible, and how each system’s interface feels in daily life, rather than dramatic differences in glycemic outcomes.

Effects on Sleep and Emotional Burden

Managing diabetes is exhausting, and a large part of what automated systems promise is not just better numbers but less mental overhead. Several studies have examined whether that promise holds. In adults with type 1 diabetes, three months on the Omnipod 5 led to significant improvements in diabetes distress, satisfaction with insulin delivery, and confidence around hypoglycemia.⁸ The psychological relief of knowing that a system is watching your glucose and adjusting insulin overnight, for example, seems to genuinely reduce the vigilance burden that wears people down over years.

The benefits were even more striking for caregivers of very young children with type 1 diabetes, a population where the burden often falls heaviest. After three months, caregivers reported significant improvements in well-being, confidence managing hypoglycemia, and sleep quality, including better sleep duration and efficiency.⁹ Parents of toddlers with diabetes often set alarms throughout the night to check blood sugars. A system that adjusts insulin automatically and alerts you only when it truly needs help can give back hours of sleep per week.

What Affects How Well the System Performs

Even the best algorithm is limited by the physics of insulin delivery. Rapid-acting insulin injected under the skin does not work instantly. It takes time to absorb into the bloodstream and then more time to exert its glucose-lowering effect. Several factors influence how quickly that absorption happens. Skin temperature plays a role: warmer skin speeds absorption, while cooler skin slows it. Subcutaneous fat thickness at the wear site also matters, with thinner tissue allowing faster insulin uptake.¹⁰

This has practical implications. If you wear the Pod on your abdomen one cycle and your upper arm the next, absorption speed can vary between the two sites. A hot shower or sauna can temporarily accelerate insulin action in ways the algorithm did not anticipate. Scar tissue from repeatedly using the same site can slow absorption unpredictably. Rotating Pod placement and being aware of temperature effects can help the algorithm’s predictions match reality more closely.

The system also cannot fully compensate for very large or very fast carbohydrate loads. If you eat a high-glycemic meal and underestimate the carb count, your glucose will spike before the algorithm can deliver enough correction insulin through basal adjustments alone. Accurate carb counting and well-timed meal boluses remain important even with automation handling the background delivery.

Cost-Effectiveness and Access

Automated insulin delivery systems are expensive upfront. Pods, CGM sensors, and insulin itself represent recurring costs that add up. Whether the investment pays off in the long run depends on how much the technology reduces expensive complications like hospitalizations, kidney disease, and cardiovascular events down the line.

A U.S.-focused cost-effectiveness analysis found that for both children and adults with type 1 diabetes, the tubeless automated system was cost-effective compared to standard care, coming in well under widely used willingness-to-pay thresholds. In more than 90% of simulated scenarios, it remained cost-effective at a $100,000 per quality-adjusted life-year benchmark.¹¹ A French analysis reached a similar conclusion, finding that although direct healthcare costs were somewhat higher with the system, savings from reduced diabetes complications and improved productivity made it cost-saving from a societal perspective.¹²

Insurance coverage in the United States varies. Many commercial insurers and Medicare now cover the Omnipod 5 for type 1 diabetes, though prior authorization requirements and out-of-pocket costs differ. Medicaid coverage is expanding but uneven across states. The real-world dataset of nearly 70,000 users included Medicaid and Medicare populations and found favorable glycemic outcomes among those groups as well, which suggests the technology works regardless of payer status when access barriers are cleared.¹

The Waste Problem

One criticism of tubeless patch pumps that rarely gets airtime in clinical discussions is environmental waste. Traditional tubed pumps have a durable pump body that lasts years; you only replace the small infusion set every few days. With the Omnipod, you discard the entire Pod, electronics and all, every 72 hours. An analysis comparing the resource loss from various insulin delivery devices found that the waste footprint of the Omnipod was orders of magnitude larger than that of conventional infusion sets.¹³ That study was published over a decade ago and assessed an earlier generation of the Pod, but the fundamental design has not changed: the Pod is disposable and contains a battery, motor, and circuit board that go into the trash with every change.

Insulet, the company behind the Omnipod, has introduced a recycling program in some markets, but participation is voluntary and the infrastructure for recovering electronics from small medical devices remains limited. For users who prioritize sustainability, the waste question is worth weighing against the quality-of-life benefits the tubeless design provides. It is one of those trade-offs where the environmental cost is real but rarely discussed because the clinical conversation focuses on glucose outcomes and user convenience.