Stroke, which is the second leading cause of death worldwide, leads to serious sequelae and has a high recurrence rate even after treatment. An international research team, including South Korea, has found a new way to manage stroke sequelae.
Professor Jeong Yoon-young from Pohang University of Science and Technology POSTECH's Department of Electrical and Electronic Engineering noted on the 27th that his research team developed a skin-mounted sensor system capable of real-time monitoring and evaluation of stroke sequelae in collaboration with the Sereneo Research Center at the Lucerne Institute in Switzerland. The results of the research were published on Jan. 7 in the international journal of digital healthcare, 'npj Digital Medicine.'
Stroke is a serious condition that occurs when blood vessels in the brain get blocked or burst, threatening life and leaving sequelae such as 'dysphagia,' which refers to difficulty swallowing, and 'dysarthria,' which leads to indistinct speech. Existing assessments of stroke sequelae have been conducted through direct examinations by medical professionals at hospitals, making it difficult to continuously track changes in patients' daily lives.
The research team developed a 'flexible skin-mounted neck vibration sensor (STVS)' to track stroke sequelae. This sensor adheres closely to the skin of the neck and is unaffected by surrounding noise, precisely detecting signals related to stroke sequelae such as speaking, swallowing, and coughing in daily life.
In particular, the research team applied a wavy structure to the sensor so that it naturally adheres to the skin and can respond to movement. The sensor can stably adhere while walking or running, allowing for continuous data measurement. Experimental results showed that this sensor improved the 'signal-to-noise ratio (SNR)' by more than three times compared to existing wearable sensors.
The research team developed an 'ensemble classification model' based on artificial intelligence (AI) to enable the automatic analysis of data collected from the sensor. Various actions related to stroke, such as swallowing, coughing, speaking, and clearing one's throat, can be accurately measured and distinguished without the help of specialized medical personnel, allowing for a high-level medical assessment. In clinical trials conducted at a Swiss stroke rehabilitation center with participants speaking five languages—Korean, English, French, German, and Spanish—the sensor developed by the research team demonstrated over 96% high accuracy in activity classification.
Professor Jeong Yoon-young said, "We have proposed a new paradigm for monitoring stroke sequelae in daily life through the integration of wearable sensors and AI technology," adding, "This technology, which has proven its high accuracy and stability in various languages and environments, will significantly contribute to the diagnosis and customized treatment of various neurological disorders in the future."
References
npj Digital Medicine (2025), DOI: https://doi.org/10.1038/s41746-024-01417-w