Abstract
Background. After a stroke, the use of the paretic arm is determined by its capacity (what it can or cannot do). When both arms have capacity to perform a task, the choice of which arm to use must be based on another criterion, probably by comparing the efficiency of each arm. Two numerical models account for this: the capacity model (the paretic arm is chosen in preference) and the efficiency model (the most efficient arm is chosen). Objective. To numerically determine whether capacity or efficiency best predict the use of the paretic arm in activities of daily living. Methods. We performed numerical simulations to predict paretic arm use with either the capacity model or the efficiency model. We used the Bayesian Information Criterion (BIC) to compare the adequacy of the 2 models in predicting clinical and accelerometric data collected from 30 patients with chronic stroke. Results. The efficiency model predicted arm use in activities of daily living better than the capacity model (BIC = -66.95 vs -5.89; root mean square error = 0.26 vs 0.72). Conclusions. The study highlights the importance of considering efficiency when assessing paretic arm non-use. Assessing individuals' arm efficiency should help personalize rehabilitation strategies after stroke.