LASSO regression for monitoring patients progress following ACL reconstruction via motion sensors: a case study

Thumbnail Image
Tedesco, Salvatore
O'Flynn, Brendan
Journal Title
Journal ISSN
Volume Title
Published Version
Research Projects
Organizational Units
Journal Issue
Inertial data can represent a rich source of clinically relevant information, which can provide details on motor assessment in subjects undertaking a rehabilitation process. Indeed, in clinical and sport settings, motor assessment is generally conducted through simple subjective measures such as a visual assessment or questionnaire given by caregivers. Thus, inertial sensor technology and associated data sets can help provide an objective and empirical measure of a patient’s progress. In this publication, several metrics in different domains have been considered and extrapolated from the three-dimensional accelerometer and angular rate data sets collected on an impaired subject with knee injury, via a wearable sensing system developed at the Tyndall National Institute. These data sets were collected for different activities performed across a number of sessions as the subject progressed through the rehabilitation process. Using these data sets and adopting a combination of techniques (LASSO, elastic net regularization, screening-based approaches, and leave-one-out cross-validation), an automated method has been defined in order to select the most suitable features which could provide accurate quantitative analysis of the improvement of the subject throughout their rehabilitation. The present work confirms that changes in motor ability can be objectively assessed via data-driven methods and that most of the alterations of interest occur on the sagittal plane and may be assessed by an accelerometer worn on the thigh.
Regression , Feature selection , Motor assessment , Rehabilitation , Wearable
Tedesco, S. and O'Flynn, B. (2019) 'LASSO regression for monitoring patients progress following ACL reconstruction via motion sensors: a case study', International Journal on Advances in Life Sciences, 11 (1&2), pp. 23-32,
2019, © Copyright by authors, Published under agreement with IARIA -