Modeling Chinese Input Interaction for Patients with Cloud Based Learning

Abstract

Parkinson's disease (PD) patients with diverse conditions face challenges in interacting with smart devices due to motor impairment. Chinese is a non-alphabetic language and text entry via smart devices highlights two major interaction challenges for motor impaired that affect the accuracy and efficiency: 1) Multiple key taps/ strokes required to enter a single Chinese character, 2) The precise selection of most relevant character from the candidate list over tiny screens. To address the first problem on hand tremor and rigidity of PD, a Chinese Input Method Editor (IME) application is designed and developed for improving the interaction and adaptability of smart devices for intended audience. To resolve the second major issue, proposed system uses cloud based learning to rank the most frequently and relevant candidate lists of Chinese characters. Auto correction feature helps in selection of with the most suitable candidate character (ranked by cloud based learning) according to the user's sentence. Further-more, a scanning technique is used to control text input via single tap over entire keypad area. Personalization feature allows PD patients with varied severity level to adapt the IME by adjusting scanning speed, and visibility (font size, back ground color) according to their pace. System has been tested with 25 PD patients to evaluate effectiveness. Our results demonstrated improved precision and ease of text input which highlights that a single touch IME solution with Cloud based learning can help PD patients to use smart devices with minimal physical fatigue and to improve their social connectivity.