Want CNET to notify you of price drops and the latest stories?

Detecting schizophrenia: The eyes have it

Researchers using video software to analyze eye movements during certain tasks said they can identify schizophrenia with "exceptional accuracy."

Elizabeth Armstrong Moore
Elizabeth Armstrong Moore is based in Portland, Oregon, and has written for Wired, The Christian Science Monitor, and public radio. Her semi-obscure hobbies include climbing, billiards, board games that take up a lot of space, and piano.
Elizabeth Armstrong Moore
3 min read

More than a century ago, doctors first realized that people with psychotic illnesses also suffer from impaired eye movements. Now researchers out of the University of Aberdeen in the U.K. have taken that observation and truly put it to the test with high resolution cameras and video software.

The result: a computer model based on a series of simple eye-tracking trials that has distinguished schizophrenics from healthy control subjects with 98.3 percent accuracy.

The eyes are the window to the psyche. Gtanner/Wikimedia Commons
"It has been known for over 100 years that individuals with psychotic illnesses have a variety of eye movement abnormalities, but until our study, using a novel battery of tests, no one thought the abnormalities were sensitive enough to be used as potential clinical diagnostic biomarkers," Philip Benson and David St. Clair wrote in the journal Biological Psychiatry.

General tests, which have been developed in recent years, include fixation, where people with schizophrenia have difficulty holding a steady gaze on a single, unmoving object, and smooth pursuit, where people with schizophrenia track slow-moving objects unsteadily, their eyes slowing down and then quickly catching up in a movement called a saccade.

For this study, the researchers recorded the eye movements of 88 people with schizophrenia and 88 healthy control subjects, and then developed algorithmic models intended to distinguish the two groups. They wrote that one model achieved "perfect separation" of the 88 schizophrenia cases from the 88 controls.

One caveat: those results come from retrospective, or backward-looking, studies in which researchers already knew which patients have been diagnosed with the disease. Diagnostic applications typically must also be tested in prospective studies against patients who haven't yet been diagnosed by traditional means.

What's more, the accuracy of the test isn't yet good enough for diagnosis. Testing a million people with a tool that's 98.3 percent accurate would falsely diagnose more than 10,000 healthy people with schizophrenia.

Still, could the eye-movement analysis someday yield an app that would diagnose the disease? Maybe. Here's what Benson had to say via email:

Eye tracking and web-based video technology is rapidly improving and becoming cheaper as well. In the near future some of the basic tests could be administered remotely. Our tests benefit from higher precision data than can't be obtained from standard webcams. We've considered smartphone and tablet applications (which have the same limitations as webcams, as well as reduced picture size and field of view), and these are of course ideal for bedside tests. With a little bit of product development we should be able to have a test that can be self-administered, with in-built quality checks.

Benson advises caution, though, stressing that the best diagnosis comes from a proper medical consultation with a specialist.

Meanwhile, he and his team are investigating whether eye movement patterns alone can also differentiate schizophrenia from other major psychological disorders such as bipolar illness. He says they already have unpublished data demonstrating that the same "battery of eye movement tests and clever statistics" can distinguish between conditions such as bipolar disorder and major depression.