Three parts that had to meet
A home closed-loop system needs three components: a sensor that measures glucose, a pump that delivers insulin, and software that decides what the pump should do. Each had its own history, and for years they developed separately. The early bedside machines of the 1970s had combined all three in one hospital instrument, but they were too large and invasive for personal use. The story of home systems is mostly the story of making each part small, reliable and safe enough to wear, and then getting them to work together.
The name “artificial pancreas” is a popular label. These systems do not replace the organ or imitate everything it does, and most only manage insulin. Researchers often prefer terms such as automated insulin delivery or closed-loop systems.
Pumps first
External insulin pumps for people with diabetes appeared in the late 1970s. Early models were large, sometimes worn like a backpack or carried in a bag, and they delivered a steady background supply of insulin with extra doses at meals. Over the next decades, pumps shrank, gained better displays, and were made easier to program. Some users adopted them, while others stayed with injections. The pump was a necessary base because a closed-loop system needs something that can adjust insulin automatically.
Continuous glucose monitoring arrives
Sensors were the long-term bottleneck. Finger-prick meters became common in the 1980s, but they give only isolated readings. A continuous monitor uses a small sensor, typically placed under the skin, that estimates glucose in the fluid around cells. In 1999, a continuous monitoring system was cleared in the United States, but it recorded data for a doctor to review afterward rather than showing it live. In the early 2000s, devices that displayed readings in real time followed, along with alarms for highs and lows. Early sensors were less accurate than finger-prick tests and often needed calibration, and a delay between blood and tissue glucose remains a challenge in the technology.
A watch-style device that tried to measure glucose through the skin without a needle came to market in the early 2000s, but it was withdrawn because of limited reliability. This episode is a reminder that not every approach worked.
Linking sensor and pump
The first step was the sensor-augmented pump, in which the sensor reading was displayed on the pump but the user still made the decisions. The next step was a safety feature: a pump that could stop insulin delivery automatically when the sensor showed that glucose was low or falling quickly. Versions of this “low glucose suspend” function reached markets in Europe around 2009 and in the United States a few years later.
Further steps added predictive features, in which the algorithm tried to anticipate a fall and reduce insulin earlier. These systems were not fully automatic, but they showed that software could safely take over part of the task.
Hybrid closed loop
A hybrid closed-loop system adjusts background insulin automatically in response to sensor readings, but the user still tells it when they are about to eat, and gives a meal dose. The word hybrid reflects that remaining human role. Research groups at universities in several countries tested prototype systems in clinical trials during the 2000s and 2010s, often starting in supervised settings such as camps or hotels and then moving to home use. Some groups also developed dual-hormone systems, which add glucagon to counter low glucose, though these have been harder to bring to the market.
In 2016, the first hybrid closed-loop system was approved in the United States for people with type 1 diabetes. Several others followed, from various manufacturers, with different algorithms and ways of working with the sensor. Their approvals were generally based on trials showing safety and improved time spent in a target range, though results varied from person to person and the systems do not eliminate the need for attention. Specific products change quickly and are not covered here.
The do-it-yourself movement
A notable thread in this history is not from any company. From about 2013, some people with diabetes and their family members, frustrated by the pace of commercial development, began building their own systems from existing pumps and sensors, using open-source software to read the sensor data and control the pump. The online slogan “We Are Not Waiting” became associated with the movement. Projects such as OpenAPS and later others shared their code and instructions freely, and communities grew around them.
These home-built systems were not approved medical devices, and their use carried risks that regulators and many doctors pointed out. Even so, the community drew attention to what users wanted, including remote data sharing and more flexible algorithms, and some of those ideas later influenced regulated products. Historians of medical technology often cite this as an unusual case of patients driving innovation. It should not be read as an endorsement of building medical devices at home.
Connected sensors and phones
Closed-loop systems also fit into the wider growth of connected sensors. Many continuous monitors send data wirelessly to smartphones, and parents or caregivers can follow a child’s readings remotely. This made use of general-purpose silicon chips and wireless radios that were developed for other markets, which helped reduce size and cost. The same trend of putting health sensors on or near the body can be seen in work on glucose monitoring lenses, although such ideas have had a mixed record.
What remains difficult
Even the best current systems face limits that engineers have worked on for decades. Insulin injected under the skin takes time to act, so the software must predict ahead. Meals, exercise, stress and illness change insulin needs unpredictably. Sensors can fail or drift, and the devices need supplies, charging and a user who can respond to alarms. Cost and access also differ widely between countries and health systems.
For these reasons, the closed-loop system is best described as a major improvement in automation, not a cure. This article is history only. Anyone considering such a device should speak with a qualified health care professional, as outcomes are uncertain and vary between individuals.