Researchers develop AI-powered stretchable computing patch

A new skin-like computing patch developed at the University of Chicago Pritzker School of Molecular Engineering (UChicago PME) can analyze health data using artificial intelligence in an unprecedented way. Unlike current wearable devices, it does not rely on wireless connectivity and performs AI calculations directly on the body in just a few milliseconds.

Current smartwatches may be able to track your heart rate and movement, but they don’t analyze what they detect. Analysis takes place elsewhere after data is transferred to an external server. In some situations, for example, detecting ventricular fibrillation in the heart, a delay of a few seconds in communicating with the server is too long.

This new device was designed and tested in collaboration with researchers at Argonne National Laboratory and was made possible by the development of a manufacturing process that allows organic electrochemical transistors to be printed on flexible surfaces.

“The future we are trying to realize is to make wearable and implantable devices smarter,” said Sihong Wang, associate professor of molecular engineering at the University of Chicago PME and co-senior author of the new study. nature electronics. “This will allow people to integrate a personal, instant doctor on their device.”

Manufacturing stretchable transistors

For years, Wang’s lab has been working to create electronic components that can stretch and bend like human skin, with the ultimate goal of creating smart devices that adhere to human tissue. The group previously developed methods for manufacturing stretchable transistor arrays and stretchable OLED displays.

In the new study, Wang and his colleagues set out to build stretchable neuromorphic computing circuits, or arrays of large numbers of transistors, that can perform analysis of health data. Previous research had demonstrated that this concept was theoretically possible with a small number of transistors, but had not been scaled up to practical sizes.

The transistors the research team wanted to use, called organic electrochemical transistors, operate differently than transistors in standard computer chips. They process information using both electrical current and the movement of ions through a gel-like electrolyte layer. Electrolytes give each transistor built-in memory, allowing it to store numbers stably over time, just as they strengthen or weaken synapses in the brain to encode learned patterns.

However, these components presented manufacturing challenges. The flexible surface layer is sensitive to heat and solvents and cannot be manufactured using standard chip manufacturing techniques. At the same time, the gel electrolyte layer tends to move like a liquid, merging with adjacent devices and causing short circuits.

“The question we had to ask was whether we could use or modify the properties of these polymers to make them compatible with photolithography, the main patterning method used in the microelectronics industry,” Wang said.

The research team solved this challenge by developing a new type of polymer gel that can be cured into precise patterns when exposed to ultraviolet light. The result is a manufacturing method that can produce 10,000 organic electrochemical transistors per square centimeter.

“As computer scientists, we are used to thinking of weights in neural networks as just numbers,” said Zixuan Zhao, a graduate student in CS at the University of Chicago and co-first author of the study. “In hardware, it’s a material with variability, history, and physical limitations. The challenge was to keep these constraints in mind and still do the calculations with sufficient accuracy.”

Save lives with fast computing

To test the new device’s usefulness, Wang’s team used one of the new stretchable arrays to run a pre-trained algorithm designed to help treat ventricular fibrillation. This dangerous electrical storm within the heart can be fatal and is most often treated with a one-size-fits-all defibrillator shock that delivers a large amount of electrical shock throughout the heart. Researchers are proposing more precise treatments. It maps abnormal electrical waves as they pass through the heart and sends out small, precise pulses just before they continue.

But time was an obstacle. The wavefront moves very quickly within the heart, so the entire analysis must be completed within milliseconds. It’s too fast to send data to an external computer and back.

“This is a situation where remote computing is not possible; it just takes too long,” Wang said. “But if we have a computing device that can analyze the inside of the body, that might be possible.”

Using real cardiac mapping data from a human donor heart, the research team showed that the stretchable array was able to determine wavefront position with 99.6% accuracy, even when the device was stretched more than 1.5 times its normal length.

In another demonstration, a neural network encoded within the array analyzed a combination of vital signs and personal health data (such as cholesterol levels, blood sugar levels, maximum heart rate, and ECG measurements) to assess a patient’s risk of heart attack, achieving 83.5% accuracy.

Wang views this computing array as one component of a fully integrated, body-compatible health platform. His lab is currently working on combining computing arrays with stretchable wireless communication components and improved sensors, working toward systems that can sense, analyze, and respond to health data as a fully integrated whole.

“Instead of sending data to a remote server, we can start to make sense of it right where life is happening,” said Fanfang Xia, a computer scientist at Argonne National Laboratory and co-senior author of the study.

Citation: “Large-scale stretchable neuromorphic circuits for on-body edge computing”, Li et al, Nature Electronics, May 20, 2026. DOI: 10.1038/s41928-026-01639-8

Funding: This research and the researchers involved were supported by the U.S. Office of Naval Research (N00014-21-1-2266, N00014-21-1-2581), the University of Chicago Joint Task Force Initiative, the National Institutes of Health (1DP2EB034563, R01-HL141470, R01 HL165002), Argonne National Laboratory, and the U.S. Department of Energy. (DE-AC02-06CH11357, DE-SC0014664), Leducq Foundation, and CZ Biohub.