Researchers at the Ulsan National Institute of Science and Technology (UNIST) have developed a next-generation "AI smart patch" that can simultaneously detect and analyze cardiovascular disease and surrounding toxic gases in real time. The device is expected to significantly contribute to protecting the safety of the elderly and patients with underlying conditions vulnerable to air pollution, as well as industrial workers operating in confined spaces.
A team led by Professor Kim Jae-jun of the Department of Electrical Engineering and Professor Jung Hoon-eui of the Department of Mechanical Engineering at UNIST said Tuesday that they had jointly developed a chest-attachable patch that integrates and analyzes biosignals such as electrocardiogram (ECG) and blood pressure together with surrounding atmospheric gas information to instantly identify diseases and hazardous situations.
The patch's most notable features are its "on-chip artificial intelligence (AI)" technology and its breakthrough "ultra-low-power design."
Existing wearable devices had to send heavy raw data to external servers or smartphones for processing. This caused communication delays and severe battery drain. In contrast, the newly developed patch analyzes data on its own inside an analog computation-based chip. Because the AI then transmits only the light final result it has determined via Bluetooth, the device minimizes communication interruptions and dramatically reduces power consumption. It is also well suited for an administrator to remotely monitor the status of multiple people at once.
The team also maximized the operating efficiency of the optical sensor (used to measure blood flow information), the main culprit of battery drain. By applying an adaptive low-power technology (RPT-PW) that turns the sensor on and off in line with the ECG signal cycle, the researchers succeeded in reducing power consumption in the sensor section by about 83% compared with existing levels. That is why the device can be worn for long periods in daily life without separate charging.
The patch also offers excellent diagnostic performance and wearability. It diagnosed hypertension or arrhythmia with high accuracy of more than 90%, and recorded 92.46% accuracy in experiments classifying toxic gas mixtures. Micro-structure technology applied to the patch's adhesive surface allows it to stay stably attached even to rough skin, and it separates easily in one direction when removed, leaving no residue.
The research is credited with simultaneously solving "power efficiency," the greatest challenge for real-time wearable devices, and the "functionality" of multi-detection. Because low-power status diagnosis is now possible on the device itself (on-device AI) without connecting to an external server, the patch has high utility both in everyday life outside hospitals and at industrial sites.
Anvix Lab, a faculty startup co-founded by the researchers, has received a transfer of the technology and is accelerating efforts to commercialize a next-generation bioelectronics patch platform based on on-chip AI.
The research findings, with UNIST Department of Electrical Engineering researchers Cho Sang-hyun and Kim Hyun-joong as first authors, are set to be formally published in the July issue of the IEEE Journal of Solid-State Circuits, the most authoritative international journal in the field of semiconductor circuit design. The research was conducted with support from the Ministry of Trade, Industry and Energy, the Ministry of Education, and the Ministry of Science and ICT.
