GMAC: A simple measure to quantify upper limb use from wrist-worn accelerometers

Abstract

AbstractVarious measures have been proposed to quantify upper-limb use through wrist-worn inertial measurement units. The two most popular traditional measures of upper-limb use – thresholded activity counts (TAC) and the gross movement (GM) score suffer from high sensitivity and specificity, respectively. We had previously proposed a hybrid version of these two measures – the GMAC – that showed better overall detection performance. However, the previously proposed GMAC used both accelerometer and gyroscope data and used the same parameter values from the TAC and GM measures. In this paper, we aim to answer two important questions to improve the usefulness of the GMAC measure: (a) can the GMAC measure be implemented using only the accelerometer data? (b) what are the optimal parameter values for the GMAC measure? We propose a modified version of the GMAC that works with only accelerometer data, and optimize this measure’s parameters. This optimized GMAC showed better detection performance than the previously proposed GMAC and surprisingly had comparable performance to that of the best-performing machine learning-based measure (random forest inter-subject model). Although intra-subject machine learning-based measures perform better than the optimized GMAC, the latter is simpler, well suited for real-time upper-limb use detection, and is the best option when a trained machine learning-based intra-subject model or labeled data is unavailable. The optimized GMAC measure can be a useful measure for either offline detection or for real-time detection and feedback of upper limb use.