Fig. 5

Classification accuracy at the start and end of trials. We observe a consistent trend of increasing classification accuracy between the start of every trial (Initial State) and the end of every trial (End State). The ultrasound images used in our KNN algorithm are again used here as our measure of the end muscle state for each hand and wrist motion. Shown classification accuracies for each hand and wrist motion are averaged across subject. Error bars denote the standard error of the mean for each hand and wrist motion. Overall Mean shows the classification accuracy of the initial and end muscle states averaged across both subject and motion