Dominate and Non-dominate Hand Prediction for Handheld Touchscreen Interaction

People have their individual preference of which hand they preferentially use to do certain things. It is not unusual to see them use mobile devices with a touchscreen one-handedly. Depending on where they are and what they do, people may use one hand over the other to hold and interact with mobile devices. Few studies have looked into the implication of using a preferred hand versus a non-preferred in touchscreen interaction on mobile devices. As the screen size increases, the difference between using a preferred hand and a non-preferred hand on the touchscreen becomes more significant. In this paper, we show how to extract features from 3 different interaction gestures on touchscreen, tap, swipe, and drag to learn if a user is using the dominant hand or the non-dominant hand. We compare the performance of using different sets of features in prediction by considering the constraints of handheld devices. A random forest-based prediction system is also created and enhanced to recognize if the user is using a preferred hand or a non-preferred hand. This technique enables the user interface of a touchscreen to adapt to which hand the user hold and interact with mobile devices.