Abstract:
The aim of this research is to improve mouse-based point-and-click target acquisition by users with cerebral palsy (CP), which is a common physical disorder among new-borns as well as youths and adults. Youths with cerebral palsy encounter difficulties accessing computers due to the spastic hand movement. A total of 46 participants with CP took part in three stages of the point-and-click experiments conducted to understand the following: 1) effect of degree of impairment on cursor control (29 participants); 2) a development of a model of movement time (MT) (the time required to travel from home object to the target object) (11 participants); and 3) testing of an algorithm to simplify these tasks (six participants). The MT showed high variability across different degrees of impairment. Therefore, it was deemed necessary to develop a model that explains the variability in MT which cannot be explained by Fitts‟s law which expresses a linear relationship between MT (dependent variable) and index of difficulty (ID) (independent variable). The developed model extends the independent variable to include a system factor, movement amplitude and human factors including erroneous click, number of sub-movements (NS), number of slip-offs (NSO), curvature index (CI), and average speed (AS). This model suggested that the CI had influenced MT variability, so therefore the CI was the basis of an algorithm including dynamic control-display (C-D) gain adjustment. The model imitates the movement described by the initial optimized impulse model; the movement is composed of two phases of sub-movements (rapid primary and slow secondary). First, a rapid primary movement is constructed towards the target. If the movement attains the target, then the task is completed. On the other hand, if the movement lands outside the target, a slower corrective movement is required. This process carries on until the target is attained. The primary goal of the task is to reach the target as fast as possible; thus, in an ideal scenario, the subject should perform a single high-velocity movement towards the target. In an ideal case, the independent variables of the developed model are only movement amplitude and AS. The algorithm aims to enhance the performance of point-and-click tasks by a dynamic adjustment of C-D gain. The algorithm showed better performance in primary submovement compared to default Windows mode (the C-D gain is set as 10) but it negated the performance in secondary sub-movements. A summary of the research contributions includes: 1) an understanding of target acquisition performance depending on degree of impairment; 2) development of software that collects information about point-and-click tasks; 3) development of a non-linear model that explains the high variability of MT during point-and-click computer tasks; and 4) an algorithm that uses dynamic C-D gain adjustments to enhance a primary sub-movement (but still needs altering to enhance secondary sub-movements).