Computers can catch mental distress in text messages, study finds

Researchers at the University of Washington have found that algorithms are as good as trained human evaluators at identifying red-flag language in texts from people with serious mental illness.

The first-of-its kind study, which looked at thousands of text messages between people with serious mental illness and their mental health providers, opens an area of study that could help both with training and scarcity of care, according to UW Medicine.

Why it matters: Texts are increasingly a part of mental health care and evaluation, but can be more difficult to assess than in-person conversations.

What they found: Computers that were programmed to evaluate texts for cognitive distortions and language suggesting a patient was overgeneralizing or catastrophizing came to similar conclusions as trained human evaluators, according to the findings published in Psychiatric Services.

• This could help detect and identify text messages reflecting distress that could slip past undertrained or overworked clinicians, the authors of the study said.

What's next: Researchers say future applications could work with wearable fitness bands or phone-based monitoring systems.