Image: The smart sensor patch is manufactured on a backing layer so that it can be peeled off and stuck to the skin (Image courtesy of Guren Matsumura, et al. Device)
Wearable sensors are devices designed to be worn on the body to measure various physiological states. As part of the Internet of Things (IoT), these sensors have great potential for health monitoring. They produce large amounts of data that must be processed to obtain a useful interpretation. The field of computing that focuses on processing this data locally on the connected sensor or device, rather than relying on a remote cloud server, is known as edge computing. This approach is essential for the advancement of wearable sensor technology. Researchers have now used edge computing on smartphones to analyze data from a flexible, multimodal wearable sensor patch to detect cardiac arrhythmias, coughs and falls.
A research team from Hokkaido University (Hokkaido, Japan) has created a flexible, multimodal wearable sensor patch and developed advanced computing software capable of identifying arrhythmia, coughing, and falls in volunteers. This innovative sensor, which uses a smartphone as an advanced computing device, is detailed in a research paper published in the journal device. The patch is equipped with sensors that monitor heart activity through an electrocardiogram (ECG), as well as respiration, skin temperature and moisture due to sweat. After ensuring their long-term usability, the sensors were embedded in a flexible layer that adhered to the skin. In addition, the sensor patch contains a Bluetooth module for connecting to a smartphone.
The team first evaluated the sensor patch’s ability to detect physiological changes in three volunteers who wore it on their chests. The patch was used to monitor vital signs in these individuals at humid globe temperatures (which assess heat stress risk) of 22°C and above 29°C. While the sample size was limited, the researchers were able to observe significant changes in vital signs while monitoring the time series at elevated temperatures. This can help identify symptoms of heat stress at an early stage. To further strengthen their findings, the team developed machine learning software to analyze the recorded data for additional symptoms, including irregular heartbeat, coughing and falls. Besides performing the analysis on a computer, they also created a sophisticated computing application for smartphones that achieved similar analytical results, with predictive accuracy exceeding 80%.
Professor Kuniharu Takei from Hokkaido University said: “Our goal in this study was to design a multi-modal sensor patch that can process and interpret data using edge computing, and detect early stages of the disease during daily life.” “The major advance of this study is the integration of multimodal flexible sensors, real-time machine learning data analytics, and remote biomonitoring using a smartphone. One drawback of our system is that training cannot be done on a smartphone, and must be done on a computer; Therefore, this problem can be solved by simplifying data processing.
