Analysis of the quality of wrist movement during a simple tracking task

Objective assessment of stroke-patients' ability to control arm movements is vital for evidence-based therapy and progress monitoring. This study compares three different indices to quantify the quality of wrist movement during a task involving tracking a target on a screen. Each method is assessed in terms of agreement between assessors (Bland and Altman limits of agreement); repeatability of readings by the same assessor (reliability coefficient); and external validity on data collected from a sample of people with impaired upper limb function and an age matched unimpaired control group. The three indices were the root mean square difference between the wrist movement and the target signal, the cross correlation between these two signals, and an estimate of the signal-to-noise ratio in the wrist movement. External validity was investigated by calculating the correlation between each measure for wrist movement, and upper limb function assessed by the action research arm test. The results of the Bland and Altman limits of agreement show that all indices were similar in performance. The cross correlation had the highest reliability coefficient for the impaired group. In terms of external validity, the cross correlation and signal-to-noise indices showed the strongest association with functional performance and may thus be the more relevant for future clinical investigations.