A claim that needs care
The phrase “first artificial pancreas” is easy to use and hard to defend. Several research groups were working on automated glucose control in the early 1970s, and they were not all working with the same goal or equipment. Most accounts mention teams in Canada, Germany and Japan, and a commercial instrument developed in the United States soon afterward. Which one counts as “first” depends on how the term is defined, so this article describes the early machines as a group rather than naming a single inventor.
What is well documented is the general design. These machines measured a patient’s blood glucose, used a calculation to decide how much insulin or glucose to give, and delivered it automatically. In control engineering this is called a feedback loop. The idea was to replace, for a short time and under supervision, the job that a healthy pancreas performs continuously.
The problem they were trying to solve
By the early 1970s, insulin had been used for roughly fifty years, but delivery was crude. Patients injected once or a few times a day, and blood sugar swung between too high and too low. Researchers wanted to learn what blood sugar and insulin would look like under tighter control, both to understand the disease and to study whether better control would reduce long-term complications. Home glucose meters, which later changed everyday care, did not yet exist in a practical form.
A machine that could track glucose continuously and respond to it would let investigators test these ideas. It was also a way to study how the body behaves in situations such as surgery, severe illness, or pregnancy, where blood sugar can be hard to manage by hand.
How the early bedside machines worked
The early systems were big, hospital-based instruments. A thin tube drew a small, steady flow of blood from a vein, mixed it with an anticoagulant to keep it from clotting, and passed it over a glucose-measuring sensor. A reading was produced every minute or so, and a controller used it to decide what to deliver.
Two pumps were typically involved. One infused insulin through a vein when glucose was high. In some designs, the other infused glucose when glucose fell too low, giving the machine a way to correct overshoots. The controller followed an algorithm that took into account both the current glucose level and how quickly it was changing. Designers experimented with how aggressive the response should be, since too strong a response could cause oscillation.
Because the sensor sat outside the body and used a flowing blood sample, the machine had to stay near the patient’s bed. The devices needed constant attention from trained staff, so they were not appropriate for everyday life. Their value was in the research clinic and the intensive care ward.
Who was involved
A few names recur in the historical record. Arnold Kadish, a California physician, is often cited for an earlier experimental closed-loop system in the 1960s that was too large and crude to be practical. In the mid-1970s, groups led by Anthony Albisser and colleagues in Toronto, and by Ernst Pfeiffer and colleagues in Germany, published work on glucose-controlled insulin infusion. Japanese researchers, among them Motoaki Shichiri, reported their own approach in the same period, including a needle-type sensor that was intended to be inserted under the skin. These efforts were largely independent, which is a common pattern in the history of inventions.
The device most people associate with the early artificial pancreas is the Biostator, a commercial instrument developed from this research and manufactured by a company in the United States in the late 1970s. It was built for hospitals and research centers rather than for sale to patients. Reports describe it as a bulky console on a cart. Researchers used it in studies of blood sugar control and in managing patients through events such as surgery and delivery.
What the machines showed
Early use of these systems demonstrated that automated control was possible. In a monitored setting, blood glucose could be held near a target for hours with less variation than manual injections generally produced. Researchers also learned a lot about how insulin and glucose interact in the body, how quickly insulin acts, and why delays are a central difficulty. Those delays remain a problem today.
They also revealed limits. Drawing blood continuously is invasive and requires anticoagulant. Sensors drifted and needed calibration. The machines were costly, hard to maintain, and not suitable for long-term use. Delivering insulin into a vein, which acts quickly, is very different from the under-the-skin delivery that most people with diabetes eventually used, where absorption is slower and more variable. These barriers meant that the bedside artificial pancreas stayed in hospitals.
From bedside to something smaller
The research did not end with the Biostator. Its main contribution was proof of concept and a body of knowledge about how to control glucose with a feedback loop. Over the following decades, researchers worked on smaller pumps, better sensors and implantable devices, and the first small insulin pumps for patient use appeared in the late 1970s and 1980s. Wearable continuous sensors were developed much later, and only then did the idea of linking sensor and pump, discussed in the story of closed-loop systems, become practical outside the hospital.
The word “artificial pancreas” continues to be used for these later systems, but the early bedside machines were a different kind of device, more like a laboratory instrument than a personal product. They are best remembered as the first demonstrations that glucose could be controlled automatically. This account is history and not a description of any current treatment, and anyone with diabetes should rely on their own medical team for care decisions.