Therapist bots: AI and mental health

When British charity The Samaritans was forced to abandon its ‘Radar’ Twitter app in 2014, many in the health community worried that emerging AI technology was poorly suited to the sensitivity of mental illness.

The app, designed to read users’ tweets for evidence of suicidal thoughts, was criticised for a host of reasons, with one online petition accusing Radar of breaching the privacy of vulnerable Twitter users by alerting everybody – friends and foe alike – of their condition.

But three years on, it appears the incident has not halted AI’s incursion into psychological healthcare, which artificial intelligence developers believe could be one of their technology’s most exciting applications.

One such backer is Jim Schwoebel, CEO of US-based NeuroLex, who made headlines last year with his tool to help doctors screen patients for schizophrenia. When Schwoebel’s brother developed psychosis, he told The Atlantic last year, doctors required more than 10 primary-care appointments before he was diagnosed.

Schwoebel’s AI service could do this job much faster, he believes, by recording a patient’s conversation with a doctor, before scanning their transcript for any linguistic clues of mental illness (his brother’s abnormal uses of the words “here” and “there”, for instance, could have indicated an approaching psychosis).

The scarcity of funding devoted to treating mental illness has become a top political issue, with Theresa May declaring earlier this year that she wanted “parity of esteem” between physical and mental health.

But with waiting times for NHS therapy remaining stubbornly long, it’s difficult to escape a central dilemma: mental health treatment is hugely time consuming, and the number of treatment-seeking patients far outstrips the number of available therapists. That’s where AI could step in to fill the void, believe many in the tech world.

AI diagnosis

Ali Parsa, founder and CEO of British app Babylon, is one such believer. He thinks the neglect of mental illness by health services across the world is “shameful”, and that, with funding from the UK government so low, AI technologies have the opportunity to “leapfrog” right into the heart of psychological treatment.

Babylon is a “virtual doctor” app, which offers patients instant, digitised consultations, as well as regular reminders to take their medicine. Earlier this year the firm announced a partnership with the NHS to introduce its AI-powered chatbot to more than 1.2 million patients in north London.

AI’s main potential within mental health, Parsa believes, lies in diagnosis. Apps like Babylon could detect whether a patient is staying at home all day – which is more likely to make them feel suicidal – by the frequency with which they check their phone, for example.

Indeed, in the world of psychological diagnosis AI is becoming ever more fashionable. US researchers from Harvard and Vermont were, by analysing colours and facial expressions, able to detect depression on popular photo-sharing app Instagram with 70% accuracy. Users with clinical depression were far more likely to post greyer and bluer photos, they found, with even the particular filter chosen by each user displaying much about their mental state.

Scientists at the University of Texas, meanwhile, are developing a series of computer exercises designed to diagnose ADHD in children.

AI and treatment

But what about when a patient has already been diagnosed? The use of AI in psychological treatment, rather than diagnosis, remains far more controversial, with many patients frightened by the near-dystopian idea of turning up at their weekly therapist session to find that their trusted doctor has been replaced by a robot.

Patients needn’t be frightened, believes Valentin Tablan, principal scientist at London-based Ieso digital health, because the use of AI in psychological treatment will never reach these extreme lengths.

Ieso – which provides online messaging therapy – has worked with more than 10,000 mentally ill patients in recent years, giving Tablan anonymised transcripts of therapy sessions totalling tens of thousands of hours. It would take years for a human to trawl through these documents and draw any meaningful conclusions, but observing correlations from vast amounts of data is what AI does best, Tablan said.

Within a few years, he said, the service should be able to give doctors much better advice around which strategies work and which don’t, and should be able to provide more personalised care – do men respond differently to women, do the young prefer different treatments to the old? With vast quantities of data at his fingertips, Tablan’s AI could prove a powerful tool.

Job replacement

So, does this mean an inevitable world of therapist bots, with the classic psychiatric “sofa chat” replaced by a faceless robot?

Not exactly. Parsa and Tablan both believe human interaction will remain an indispensable part of psychological treatment.

“It’s important to have human interaction,” said Tablan, “that’s something computers find difficult to replicate.”

The UK government does not have a perfect track record for investing in new health technology, with the Ministry of Defence, for instance, declining to fund an award-winning virtual reality treatment aimed at veterans suffering from PTSD.

But it is now impossible to ignore the growing political demand for increased mental health funding. The UK’s Liberal Democrat party made the issue a central feature of its policy manifesto at the last General Election, and health experts from across the board have come out demanding parity of funding between physical and mental health.

With doctor time in increasingly tight demand, it seems likely that AI “doctor bots” may become the chosen path.