For those living with locked-in syndrome, cerebral palsy, and other conditions that cause motor impairment, technology and software that enable communication have come a long way. Empowering those who can’t talk to communicate with computer interfaces has changed lives for the better, but some of these interfaces can still be frustratingly slow for the users.
Researchers at MIT are developing an interface system that would greatly improve speed and accuracy for individuals communicating with this technology.
Current communication interfaces use a row-scanning system where the user selects one letter or word at a time by activating a switch, blinking, or blowing a puff of air. It’s understandable how having a conversation by spelling out each word this way can be frustrating.
The new system from MIT, called Nomon, uses probabilistic reasoning to learn how users make selections and adjusts to optimize speed and accuracy.
Participants who tested the system found that they could type and communicate much faster using Nomon than they did with more typical row-column scanning systems. The participants also did better on a task where they selected pictures, which shows that Nomon could be used for more than typing.
Nomon uses machine-learning algorithms to learn what users select over time, identify patterns, and then personalize their available selections. The system very much becomes the unique communication tool of the user. It not only predicts things like spelling, much like autocorrect, it also learns where the user typically clicks on a selection box and can thereby correct when a user makes a mistake. It also puts screen selections in personally intuitive places that can go anywhere on the screen.
“Nomon is much more forgiving than row-column scanning. With row-column scanning, even if you are just slightly off, now you’ve chosen B instead of A and that’s an error,” says senior author Tamara Broderick, an associate professor in the MIT Department of Electrical Engineering and Computer Science.
In their studies with 13 users with an advanced form of spinal muscular dystrophy, participants were able to type 20 percent faster overall and made fewer errors. The team plans to further test Nomon and fine tune it more, planning to partner with more motor-impaired individuals.
“It is so cool and exciting to be able to develop software that has the potential to really help people. Being able to find those signals and turn them into communication as we are used to it is a really interesting problem.”