Multi-label classification for the analysis of human motion quality

Abstract

Knowing how well an activity is performed is important for home rehabilitation. We would like to not only know if a motion is being performed correctly, but also in what way the motion is incorrect so that we may provide feedback to the user. This paper describes methods for assessing human motion quality using body-worn tri-axial accelerometers and gyroscopes. We use multi-label classifiers to detect subtle errors in exercise performances of eight individuals with knee osteoarthritis, a degenerative disease of the cartilage. We present results obtained using various machine learning methods with decision tree base classifiers. The classifier can detect classes in multi-label data with 75% sensitivity, 90% specificity and 80% accuracy. The methods presented here form the basis for an at-home rehabilitation device that will recognize errors in patient exercise performance, provide appropriate feedback on the performance, and motivate the patient to continue the prescribed regimen.

Evaluating squat performance with a single inertial measurement unit

“…The classification results presented in table III are comparable to this work while having the added benefit of using only one IMU. Taylor and colleagues [16] built on these results in 2012 using a number of different methods to classify human motion quality. Their highest accuracy for the standing hamstring curl was 86% and for the straight leg raise was 90%.…”

“…More recent research has also shown that it may be possible to evaluate these exercises more comprehensively [3], [15], [16], and possibly with a single IMU [7], [17]. This study differs from previous work in the field as it aims to evaluate if a single body-worn IMU is capable of distinguishing between seven levels of performance in a compound exercise (i.e.…”

“…The same authors built on this work in 2012 [16] and evaluated the use of multi-label classifiers to assess exercise performance in patients with knee osteoarthritis using five IMUs. On this occasion each IMU contained a tri-axial gyroscope as well as an accelerometer.…”

Proposed Design Approach for an Interactive Feedback Technology Support in Rehabilitation

“…Many researchers have developed IMU-based systems to support patients with orthopaedic or musculoskeletal conditions as they perform their rehabilitation exercises at home, providing them with real-time feedback [20][21][22][23][24]. However these platforms have only been evaluated in laboratory based studies.…”

“…The results of this investigation showed that the 2-week interactive rehabilitation programme was as effective as the conventional rehabilitation programme, thus highlighting the promise for a sensor-based rehabilitation support. However it is worth noting that this system, as well as those referenced in [20][21][22][23][24], used multiple sensors to quantify exercise performance. Using multiple sensors is not convenient for patients to set-up and use in the home.…”