Researchers at Washington State University (WSU) in the US have developed an algorithm tailored to improve the accuracy of wearable health devices.
The algorithm claims to compensate for missing health data from sensors, potentially to benefit users in remote or underserved areas.
WSU stated that wearables are gaining traction for their ability to monitor vital signs and assist in healthcare, particularly in rural regions.
They rely on sensors and machine learning algorithms to provide health insights. However, missing or incomplete data due to user error, energy constraints, or sensor malfunctions can lead to inaccuracies.
WSU's School of Electrical Engineering and Computer Science’s Raymond and Beverly Lorenz distinguished assistant professor Ganapati Bhat spearheaded the research.
Bhat explains that although machine learning algorithms are designed with the assumption that data from all sensors will be available, this is frequently not the case.
Bhat said: “Missing data can lead to a significant drop in performance of the health algorithms. In the worst case, it can miss catastrophic cases like falls, which impact user health.
“The key insight is that we do not need the exact representation of the missing sensor data if we can maintain high predictive accuracy for the health task.”
Bhat's National Science Foundation CAREER award provided the funding for the work.
The WSU team's approach to representing missing data aims to maintain accuracy while being energy-efficient, a critical consideration for battery-powered wearable devices.
The team includes graduate students Taha Belkhouja and Dina Hussein, as well as associate professor Jana Doppa.
Their method was validated in various wearable health applications, such as assistive devices for usage in paralysed individuals, and revealed to be accurate even in the absence of multiple sensors.
The team is currently looking to collaborate with the WSU School of Medicine to test their algorithm in real-world scenarios involving gesture and activity identification.
