01 Sep 2018

Your Keyboard Is Listening

Can your typing habits predict brain disease?
by April White


Illustration by Chris Gash

For his latest venture, serial entrepreneur Jan Samzelius (MBA 1979) set out to eliminate one of the biggest annoyances of the digital era: hard-to-remember passwords. But along the way, the data security expert stumbled on a tool with the potential to change how doctors diagnose and treat neurological diseases such as Parkinson’s, Huntington’s, and Alzheimer’s.

Samzelius and his team at NeuraMetrix, based in the Bay Area, had been studying typing cadence, measuring how long subjects pause between two particular key strikes or hold down a certain key, collecting more than 2,000 data points. Research has shown that this behavioral biometric is highly consistent over time, seemingly unique to each individual typist, and nearly impossible for another person to replicate. Samzelius believed that monitoring these typing patterns could allow for continuous verification of a computer’s user, eliminating the need for passwords. He was making just that pitch in a meeting in 2013 when someone offhandedly said, “Hey, I bet this could also be used to detect Alzheimer’s.”

“This is what we in the area of neurodegenerative diseases have been looking for for many, many years,” says Bob Mahley, president emeritus of the Gladstone Institute of Neurological Disease. Samzelius, who had no knowledge of neurological diseases, first sought out Mahley for advice shortly after that fateful meeting. (Mahley now serves as chairman of NeuraMetrix’s scientific advisory board.) When he heard Samzelius’s ideas, Mahley recognized that many of the characteristics that made typing cadence a promising data security solution also made it a promising diagnostic tool because of its reliability and because it can be measured inexpensively and frequently. And unlike many other medical tests used to diagnose neurological diseases, it is also noninvasive and therefore can be used to monitor healthy people for signs of disease.

“A lot of people may say, ‘Do I really want to know if I have one of these diseases?’”

“A lot of people may say, ‘Do I really want to know if I have one of these diseases?’”

“Researchers think there are subtle signs that precede the known symptoms of diseases like Huntington’s and Alzheimer’s,” explains Robi Blumenstein (MBA 1984), president of the CHDI Foundation, a nonprofit that funds and manages Huntington’s disease research. “If we could measure those signs accurately, we could treat these diseases earlier, before significant damage has taken place in the brain.” NeuraMetrix’s early studies have shown it can detect Parkinson’s, says Samzelius.

To use it, a person installs NeuraMetrix software on his or her computer to monitor typing patterns. (Data on what is typed is not recorded.) Then, each time the person uses that computer, information is uploaded to NeuraMetrix, where it is analyzed for inconsistencies. Because the amount of data collected is immense and the algorithms are looking for changes in patterns, not changes in speed, being tired or distracted while typing will not set off false alarms. An easy-to-read dashboard shows the results: A healthy individual will see little inconsistency in cadence, while those with Parkinson’s, for instance, will see significant variation in typing patterns.

NeuraMetrix is also studying the potential of using typing cadence to diagnose or monitor ADHD, depression, and REM sleep behavior disorder, with more trials planned. The next step is commercialization. They have begun their efforts in Europe, Samzelius says, where single-payer health plans simplify negotiations, and plan to market the product as a tool for monitoring diagnosed patients for changes brought on by disease progression or treatment regimens. The bigger regulatory hurdle—and the even bigger market—is in early detection and the “worried well,” those healthy individuals who are concerned about developing neurological diseases, he says. “A lot of people may say, ‘Do I really want to know if I have one of these diseases?’ Well, research shows that, yes, people do want to know. Because then they can plan.”

Featured Alumni

Featured Alumni

Class of MBA 1979, Section I
Class of MBA 1984, Section A

Post a Comment