Clinical observations on hemiparetic patients have recorded that most patients underutilise their residual motor capacity in natural settings [1]. These patients use their less-affected arm to carry out most activities of daily living (ADL) and do not maximally utilise the residual motor capacity of their more-affected limb. This asymmetry in arm use could cause secondary complications and lead to poor quality of life. This thesis aims to (i) assess the arm use pattern of healthy and hemiparetic populations in their natural setting and (ii) understand how different training regimes affect arm use asymmetry through a planar reaching experiment with healthy young adults.
Wearable sensors provide continuous, unobtrusive, and objective assessment of arm movements in natural settings. A clinical trial was performed to measure arm movements using a wrist-worn IMU device and characterise gross movement score (GM) developed by Leuenberger et al. – a metric to measure functional arm use [2]. The characterisation study showed that the GM score has poor sensitivity but high specificity in detecting functional arm movements. The in-clinic characterisation study followed a 7-day at-home feasibility evaluation to use two wrist-worn IMU devices to measure arm use asymmetry. The data from the at-home study was used to develop a visualisation method and quantitative measures of arm-use asymmetry. Based on the insights from the two clinical trials, a framework for sensor-based assessment of upper limb functioning of hemiparetic patients was developed. The framework included (i) definitions and visualisations of important constructs associated with upper-limb functioning, (ii) a measure to quantify how much the upper-limb is used, and (iii) a measure to quantify the asymmetry in using them.
In a choice-based planar reaching experiment with healthy young adults, two popular training regimes – constraint-based and reinforcement-based training were compared to better understand the factors contributing to arm use asymmetry and how each regime helps reverse it. It was observed that both training regimes helped participants adapt to the rotation, and there was no significant difference between the arm choice of the two groups at the group level. However, at the individual level, high variability in participants' arm choice behaviour was found. Exploratory cluster analysis showed 4 different types of responses to training, indicating that a one-size-fits-all rehabilitation method might not work effectively. These results show that personality traits might be worth considering during therapy selection for reversal of arm non-use.