Organic electrochemical transistors (OECTs) are neuromorphic transistors made of carbon-based materials that combine both electronic and ionic charge carriers. These transistors could be particularly effective solutions for amplifying and switching electronic signals in devices designed to be placed on the human skin, such as smart watches, trackers that monitor physiological signals and other wearable technologies.

In contrast with conventional neuromorphic transistors, OECTs could operate reliably in wet or humid environments, which would be highly advantageous for both medical and wearable devices. Despite their potential, most existing OECTs are based on stiff materials, which can reduce the comfort of wearables and thus hinder their large-scale deployment.

Researchers at the University of Hong Kong have developed a new wearable device based on stretchable OECTs that can both perform computations and collect signals from the surrounding environment. Their proposed system, presented in a paper published in Nature Electronics, could be used to realize in-sensor edge computing on a flexible wearable device that is comfortable for users.

"The rise of AI and machine learning has been transformative, permeating various fields," Shiming Zhang, co-author of the paper, told Tech Xplore.

"However, their deployment in wearables, which is crucial to enabling digital health, is just beginning. Our goal is to embed machine learning capabilities into wearables to enable in-sensor neuromorphic computing or edge-computing capabilities. This allows for real-time, edge-based decision-making, which is vital for closed-loop theranostics and relevant to AI-driven medicine."

As part of their study, Zhang and his colleagues set out to develop an AI-powered wearable device that is based on stretchable OECT arrays. This firstly entailed developing machine learning algorithms and training them on bio-medical datasets to accurately make specific predictions about the physiology and health of users.

"To merge our algorithms with wearables, we face three main challenges: collecting higher-quality health data for precise training, suppressing motion artifacts with soft skin and thus minimize data noise, and customizing an algorithm for maximum computational efficiency," said Zhang.

"Correspondingly, we use OECTs to achieve high-quality muscle EMG signals; develop stretchable OECTs to minimize motion artifacts, and employ a specific AI algorithm, reservoir computing, for energy-efficient data training."

The OECT fabricated by the researchers and integrated in their proposed wearable device are made of stretchable components, including an elastomeric substrate, a semiconducting polymer-based channel and a solid-gel electrolyte, as well as gold-based source, drain and gate electrodes. The transistors were found to exhibit a stretchability of over 50%, reaching sizes down to 100 μm.

The researchers fabricated their stretchable transistors using a high-resolution inkjet printing system and subsequently used them to develop a smartwatch-compatible in-sensor computing module. In initial tests, this module was found to perform remarkably well, for instance, predicting the hand gestures of users wearing it with an accuracy of approximately 90%.

"In this project, we synergize multidisciplinary knowledge—spanning materials science, manufacturing, electronics, AI, and medicine," added Zhang.

"The presented WISE platform (Wearable, Intelligent, and Soft Electronics) is universal and can be easily customized for other computational wearable applications. This system has the potential to improve health outcomes for a wide range of diseases, benefiting both patients and the broader public."

© 2024 Science X Network