Keyboards are smart things these days. Some use machine learning to speed up our typing by predicting which letter or word we’ll want next. Others look at our typing habits and try to use this to guess our emotional state. Now, researchers from Australia’s Charles Sturt University have a new idea: Using the subtle clues in a person’s typing to look for early signs of Parkinson’s disease.
Parkinson’s is a progressive neurodegenerative brain disorder, which causes a range of physical and psychological symptoms. More advanced Parkinson’s is noticeable through physical traits, such as involuntary tremors and the distinctive shuffling walk known as Parkinsonian gait. However, it can be tough to spot early on before such symptoms have manifested.
That’s a problem because this period before diagnosis is a crucial time for patients. By the point of diagnosis, 70 percent of the brain’s dopaminergic neurons — the neurons which synthesize the neurotransmitter dopamine — may have already been irreversibly lost. While there is currently no cure for Parkinson’s, there is plenty of active research into finding drugs that will slow or halt its progression. For these drugs to be effective, early intervention is of the utmost importance.
The Charles Sturt researchers have developed software that monitors the time between key presses on a keyboard and uses this as a measure of hand tremor frequency. In tests, the system was able to correctly identify patients with mild Parkinson’s with 78 percent accuracy. While that’s not perfect accuracy by any means, it’s still enough that a person might want to get themselves examined more closely.
“Around 75 percent of PD sufferers have hand tremor as one of their symptoms, so being able to detect that tremor from keystroke characteristics can provide a second cardinal feature to satisfy the clinical diagnosis of the disease,” lead author Warwick Adams told Digital Trends. “What makes this finding so exciting is that it can lead to development of a widely available screening test for early PD: One that will be able to be used by first-level clinicians, as well as by individuals themselves.”
In addition to aiding with diagnosis, the technology could also help monitor the effectiveness of individual medication. Even without a Parkinson’s cure, this could be useful right now, since the required dosage of dopamine replacement medication changes over time.
“The reason for focusing on typing is that it is something that nearly everyone does, and it can be monitored as people use their own computer in their own home, without needing any specialized equipment,” Adams continued. “There is already a patent pending and, over the next couple years, the plan is to commercialize this research — ideally in partnership with a drug company or diagnostic services provider.”