- βFDA-cleared closed loop systems increase time-in-range by an average of 2+ hours per day based on pivotal trial data
- βSignificant reductions in nocturnal hypoglycemia demonstrated across multiple real-world studies
- βThree commercially available systems as of 2025 with ongoing development of fully closed loop options requiring no meal announcements
What Is a Closed Loop System?
A closed loop insulin delivery system β often called an "artificial pancreas" β is a three-part technology that continuously measures blood glucose, calculates the right insulin dose, and delivers it automatically, without the person with diabetes having to manually intervene for most decisions. The name comes from the idea of "closing the loop" between glucose sensing and insulin delivery, a loop that people with diabetes have historically had to close themselves with every meal bolus and correction dose.
Understanding how this works at a technical level helps patients make informed decisions about whether current systems β or trials of next-generation systems β are right for them.
The Three Components: How They Work Together
1. The Continuous Glucose Monitor (CGM)
The CGM reads interstitial glucose levels (the fluid just under the skin) every 1β5 minutes and transmits data wirelessly. Current FDA-cleared CGMs used in closed loop systems include the Dexcom G6, Dexcom G7, and Medtronic Guardian 4 sensor. The accuracy of this data is foundational β a reading that's 15β20 mg/dL off can cause the algorithm to make suboptimal decisions.
2. The Control Algorithm
This is the brain of the system. The algorithm receives CGM data and predicts where glucose is heading over the next 30β90 minutes using mathematical modeling. Most commercial systems use a form of Model Predictive Control (MPC) β the same type of algorithm used in industrial process control and autonomous vehicles. It essentially asks: "Given where glucose is now, where it's been, and what insulin is already working in the body (called insulin on board, or IOB), what delivery rate in the next 5 minutes will keep glucose in range?"
The algorithm runs continuously. Every 5 minutes it gets new CGM data, recalculates, and adjusts the basal rate up or down, or suspends delivery entirely if hypoglycemia is predicted. Some systems, like the Omnipod 5 running the Horizon algorithm, can also issue automatic correction boluses β not just basal adjustments.
3. The Insulin Pump
The pump receives commands from the algorithm and delivers micro-doses of rapid-acting insulin through a cannula under the skin. Modern pumps communicate with the algorithm via Bluetooth in near real-time. The physical limitation here matters: insulin injected subcutaneously takes 15β20 minutes to begin working and 2β4 hours to fully clear. This physiological delay is the primary challenge all closed loop systems must compensate for.
What the Clinical Evidence Shows
Closed loop technology has moved well beyond proof-of-concept. The pivotal trials supporting current FDA-cleared systems produced real, measurable outcomes:
- The DCLP3 trial (Control-IQ, Tandem Diabetes Care) β published in the New England Journal of Medicine in 2019 β showed that patients using Control-IQ spent 71% of time in the target glucose range (70β180 mg/dL), compared to 59% in the control group, with significant reductions in hypoglycemia.
- The FLAIR trial compared faster-acting insulin aspart to standard aspart in closed loop systems, finding modest improvements in time-in-range with the faster formulation β highlighting that the speed of insulin action still matters even when a computer is doing the dosing.
- The Omnipod 5 pivotal trial demonstrated an average time-in-range improvement of approximately 2.4 hours per day in adults and children with Type 1 diabetes compared to their previous therapy.
These are not small or theoretical gains. Each additional hour spent in glucose range per day is associated with reduced risk of long-term complications.
Current Limitations: What Closed Loop Can't Do (Yet)
Current systems still require users to announce meals β entering a carbohydrate estimate so the algorithm can pre-bolus. This is the most significant remaining manual step, because the physiological delay of subcutaneous insulin means the algorithm can't react fast enough to a meal it doesn't know is coming. Research into "fully closed loop" systems that eliminate meal announcements is ongoing, but none have yet received FDA approval for general use.
Other real-world limitations include sensor failures, infusion set occlusions, and the fact that algorithms are tuned for typical physiology β they can struggle with unusual insulin resistance patterns, high-fat meals, or exercise.
Current Status (2025)
As of 2025, three closed loop systems are FDA-cleared and commercially available in the United States: Tandem Control-IQ, Omnipod 5 (Insulet Corporation), and Medtronic MiniMed 780G. All three are indicated for Type 1 diabetes; the 780G also carries an indication for Type 2 diabetes in adults requiring intensive insulin therapy. Access through insurance remains inconsistent, and out-of-pocket costs can be a barrier. Patients can explore compatible supplies and system components through resources like mdsdiabetes.com.
Next-generation systems in active development include iLet Bionic Pancreas from Beta Bionics (bihormonal dosing of both insulin and glucagon) and fully closed loop research trials β many of which can be found at ClinicalTrials.gov by searching "closed loop insulin delivery" or "artificial pancreas."
What This Means for Patients
For people with Type 1 diabetes, current closed loop systems represent the most significant advance in daily management since the CGM itself. Real-world data show meaningful improvements in time-in-range and dramatic reductions in nocturnal hypoglycemia for most users. They are not a cure, they require ongoing engagement, and they work best when users understand the system's logic rather than treating it as a black box. Asking your endocrinologist specifically which system is compatible with your current CGM and pump is the right first question. If you don't currently use an insulin pump, the transition requires training and a prescription β but the outcomes data increasingly supports making that transition.
