ETH Zurich built a wireless device that converts glucose to gluconate on a gold-and-platinum surface, switching itself on when a sensor says blood sugar is too high.

Every treatment for type 2 diabetes works by instructing the body. Insulin tells cells to take up glucose; metformin tells the liver to make less of it; the newer incretin drugs tell the pancreas and the brain to behave differently. The common feature is that the body remains the thing doing the work.

A team at ETH Zurich has built something that does the work itself. Writing in Science, they describe an implantable device that chemically converts glucose into gluconate on a catalytic surface, removing it from circulation without any cells, enzymes or drugs involved.

“For decades, we have treated diabetes by telling the body how to manage glucose,” said Martin Fussenegger, a professor in ETH Zurich’s Department of Biosystems Science and Engineering in Basel, who led the work. “We are now showing that we can take control of glucose directly, using electronics rather than drugs.”

What the catalyst is

The active element is graphite felt carrying gold nanoclusters, which are themselves decorated with smaller platinum particles. Applying a voltage drives the oxidation of glucose to gluconate, which the body then excretes in urine.

Using an entirely inorganic catalyst is the design decision that distinguishes this from the field’s usual approach. Fussenegger’s laboratory is better known for engineered-cell implants, and a catalyst has advantages a living implant does not: nothing to keep alive, nothing to reject, no risk of the cells doing something unintended, and a shelf life governed by chemistry rather than biology.

Gluconate is a reasonable product to aim for. It is a benign compound, already present in the body and used widely as a food additive and a mineral carrier, and the kidneys clear it. Converting a sugar the patient cannot process into something inert that leaves in urine is a route that does not depend on the metabolism that is failing.

Closing the loop

The system is designed to run itself. A standard glucose sensor monitors tissue glucose, sends a signal by Bluetooth to a smartphone or smartwatch, and the phone activates the catalyst. Once glucose reaches the target level, the device switches off.

“And the catalyst only activates when glucose levels are too high,” Fussenegger said.

That is the part that makes it plausible as a therapy rather than a hazard. A catalyst running continuously would keep consuming glucose below the safe range, and hypoglycaemia is the risk that constrains every aggressive diabetes treatment — it is what makes insulin dangerous and what artificial pancreas systems spend most of their engineering effort avoiding.

Power is delivered wirelessly at 13.56 megahertz, a frequency reserved for industrial, scientific and medical use, and the electronic control unit sits in a 3D-printed casing. Wireless power means no implanted battery, which removes the commonest reason implanted devices need replacing.

What has been shown

Two kinds of test are reported. In the laboratory, the catalyst converted glucose in a five-litre reservoir — roughly the volume of blood in an adult human, which is the comparison the choice of volume is making.

In diabetic mice, blood sugar was regulated precisely, the animals lost weight, the implant was well tolerated, and it operated reliably over six weeks.

Safety was addressed directly, since passing current through tissue is the obvious concern. “The system operates at a very low voltage, and our experiments showed no evidence of tissue damage from either electrical or thermal effects,” Fussenegger said.

Six weeks of reliable operation in a mouse is a meaningful durability result for a catalytic implant, where the failure modes are fouling of the catalyst surface by proteins and cells, and encapsulation of the device in scar tissue that cuts it off from the fluid it is meant to treat.

The distance to a patient

The device has not been tested in humans, and the gap is wide enough to be worth itemising.

Scale is the first issue. A mouse holds a couple of millilitres of blood and a human around five litres, and the glucose load to be cleared scales accordingly. The five-litre reservoir test addresses this in principle, but a benchtop reservoir is stirred, clean and free of the proteins, cells and competing molecules that an implant actually sits in.

Selectivity is the second. A catalyst that oxidises glucose at low voltage is unlikely to oxidise only glucose, and blood contains a great many other oxidisable species. The release does not report what else the surface converts, or what those products are.

The third is the loop itself. A closed-loop system that depends on a continuous glucose sensor inherits that sensor’s lag and drift, and one that routes its control signal through a consumer smartphone inherits whatever happens when the phone’s battery dies or the Bluetooth connection drops. Artificial pancreas systems have spent years on exactly these failure modes.

The paper’s own framing — “drug-free management of diabesity” — describes an ambition rather than an available option, and nobody with diabetes should read a preclinical mouse result as bearing on their own treatment.

What the work does establish is that the approach functions: an inorganic catalyst can lower blood glucose in a living animal, under automatic control, without damaging the tissue around it. That is a different category of intervention from anything currently used, and the first demonstration that it works at all is the step that had to come first.

The release names no other authors and does not state the study’s funding sources.

---

Notes

1. “Wireless electrocatalytic closed-loop system for drug-free management of diabesity.” Science, 2026. DOI: 10.1126/science.ady8560.
2. ETH Zurich. “Tackling diabetes with an electronic implant: how a tiny catalyst lowers blood sugar.” EurekAlert!, 22 October 2026. https://www.eurekalert.org/news-releases/1148869.