[ALS and communication aid--a linguistic approach to scanning input]

Supporting communication disorders is one of the most important tasks that Information Technology is imposed. In this commentary, recent progress in improving efficiency of the scanning communicator, on which considerable number of amyotrophic lateral sclerosis (ALS) patients are relying, is presented. Scanning is a human interface for selecting items such as commands or characters. Scanning communication aids are widely used by a number of physically-challenged people. However, it has a serious disadvantage of its time-consuming input operation. The author's group has developed a novel scanning input method, which is called pairwise scanning. It adopts multiple cursors for accelerating character selection. A sequence optimization algorithm with a statistical language model finds the most likely combination of characters that the user is expected to input. Assuming 95% precision in the estimation, proposed method can accelerate the communication throughput up to 30%. The author's another approach is the automatic error correction in scanning input based on statistical language model. The key idea is to model the individual user's motor characteristics by examining her/his switch timing distribution, which yields another statistical model called switch timing model. A dramatic improvement to the accuracy is obtained by introducing the model. As a result, proposed method improved the character correct rate from 77.7% to 97.7% for an ALS patient.