Reconstruction of Handwriting Traces

Examples of reconstructed traces from one recording session. Actual traces are shown in blue; reconstructed traces are shown in red.

Neuroscientists have developed a fingerless glove that automatically translates hand motions into text by way of electrode sensors. Michael Linderman and his colleagues published the results of the first phase of their research project in last Wednesday’s PLoS ONE. In this phase, six volunteers, using a digital pen, wrote the numerals 0 to 9 fifty times while wearing the prototype glove, which recorded the electrical activity of eight muscles in their hand and forearms. A computer correlated the electrical data with the output from the digital pen, and pattern recognition taught the computer to derive written symbols from the bursts of electrical activity.

Linderman’s team also used a technique called discriminate analysis to test how well the computer could recognize muscle-movement patterns as corresponding with particular written characters. The computer recognized 63 percent of numerals that were repeated five times, and 97 percent of the numberals that were repeated 35 times.

Now, as much as it might be nice to stop and take notes any where at any time… can’t we already do that? I’m not totally convinced that donning a special glove is any more convenient (let alone cost-effective) than whipping out a pen and paper. Linderman and his team, however, are developing this technology with its potential medical applications in mind. Many neuromuscular disorders, like Parkinson’s and Alzheimer’s, include hand tremors as first symptoms, and Linderman thinks the glove could potentially catch these symptoms at an earlier stage than current methods do. He also thinks it might be useful in helping patients with hand tremors learn to write again, or in creating prosthetic writing devices. The details of just how it improves upon a pen or pencil for these purposes remain to be divulged.