Estimation of human trunk movements by wearable strain sensors and improvement of sensor's placement on intelligent biomedical clothes

Abstract BACKGROUND: The aim of this study was to evaluate the concept of a wearable device and, specifically: 1) to design and implement analysis procedures to extract clinically relevant information from data recorded using the wearable system; 2) to evaluate the design and placement of the strain sensors. METHODS: Different kinds of trunk movements performed by a healthy subject were acquired as a comprehensive data set of 639 multivariate time series and off-line analyzed. The space of multivariate signals recorded by the strain sensors was reduced by means of Principal Components Analysis, and compared with the univariate angles contemporaneously measured by an inertial sensor. RESULTS: Very high correlation between the two kinds of signals showed the usefulness of the garment for the quantification of the movements' range of motion that caused at least one strain sensor to lengthen or shorten accordingly. The repeatability of signals was also studied. The layout of a next garment prototype was designed, with additional strain sensors placed across the front and hips, able to monitor a wider set of trunk motor tasks. CONCLUSIONS: The proposed technologies and methods would offer a low-cost and unobtrusive approach to trunk motor rehabilitation