Predicting stroke gesture input performance for users with motor impairments

The performance of users with motor impairments with stroke gesture input on touchscreens has been little examined so far, despite the wide prevalence of mobile devices and the benefits they bring to increase users' quality of life. In this work, we present the first empirical results on this subject matter from 915 gestures collected from 10 participants with motor impairments (spastic tetraplegia and tetraparesis) and 10 participants without known impairments. We report that different motor abilities lead to different performance in terms of gesture production time. We also show that the production times of gestures articulated by users with motor impairments can be accurately predicted with an absolute error of just 150 ms and a relative error of only 3.7% with respect to actual times (user-independent tests), a result that will enable designers to estimate human performance a priori when prototyping gesture UIs for users with motor impairments.