MEAGHAN FRITZ used to break out in cold sweats before meetings. A simple discussion with her boss would leave this otherwise confident woman stuttering and dizzy. To meet her, you’d never imagine she suffered from severe social anxiety. Then again, it surprised her too. Fritz had been treated for depression for years. “I’ve tried Seroquel, Lexapro and Xanax,” she says, rattling off a list like someone who has tried it all. But nothing quite worked. In fact, according to Fritz, she wasn’t even looking for help any more. “I knew I had a problem, but I didn’t know what it was,” she says. “It wasn’t until I took the test at Joyable that I had a name for it.”
Joyable is a start-up that offers counselling over the internet. It not only finally gave Fritz her diagnosis but, $100 and 12 online sessions later, she was seeing improvements in her private and professional life too – all without meeting a therapist. Joyable is part of a new generation of computer-based therapies that some think could greatly reduce the burden on mental healthcare providers – and perhaps do away with face-to-face sessions altogether.
Advertisement
Read more: Brain and mental health
A new collection of some of the best, recent New Scientist articles on mental and neurological health to highlight World 91ɫƬ Day
“There’s a genuine need for something that can be used to meet the unmet needs of people with depression,” says at the University of York, UK, who studies mental health services. “There’s still overprescription of antidepressants and woefully inadequate provision of psychotherapies.” As resource-strapped organisations like the UK’s National 91ɫƬ Service struggle to meet the demand for treatment, online tech is taking up the slack. The . Its – a pilot project that ran from March 2013 until this summer, now being revised – approved 26. And there are hundreds more, like Joyable, without official backing. Some offer do-it-yourself counselling, some put people in touch with healthcare professionals remotely, and some are built on peer support.
There is little doubt that computer-based treatments will play a big part in the future of mental healthcare. The UK government recently launched a £650,000 innovation prize to encourage the development of the next generation of tech. But is it the solution or are we offloading care by uploading it?
Cognitive behavioural therapy (CBT) is one of the most widespread treatments for mental health conditions such as anxiety and depression. By talking with a therapist, patients are helped to change the way they think and behave. The idea behind computer-based CBT (CCBT) is to achieve similar results by delivering the change-inducing message on screen rather than in person.
“CBT relies much more on educating people than on the relationship that develops between a person and the therapist,” says psychiatric epidemiologist Glyn Lewis at University College London. “So the main online applications have tried to capture that educational element.”
There are clear advantages. For a start, computer-based services are available immediately. In the UK, half of those who need mental healthcare have to wait three months for any other kind of treatment; 1 in 10 have to wait over a year. Even for patients receiving therapy, a computer program is arguably far more convenient. In practice, psychotherapy can be a hassle, says Lewis. “You’ve got to turn up somewhere every week, you often have to miss work.” And then there are those who would prefer not to seek help from a therapist at all because of the stigma still attached to mental health problems. Computer-based alternatives allow more flexibility in how CBT is delivered.
The trouble is, they might not work. True, a handful of recent studies concluded that CCBT often gave results as good as – and sometimes better than – face-to-face therapy. There was enough evidence for the UK National Institute for 91ɫƬ and Care Excellence (NICE) to tentatively recommend the use of online apps in its current guidelines, says Gilbody.
Treating depression: Deep brain stimulation – A wonder treatment pushed too far?
But Gilbody and colleagues are about to publish results that overturn those findings. In the largest trial so far, the team recruited 690 patients who were being treated for depression across the UK. It looked at whether either of two CCBT packages – Beating the Blues, a market leader developed by London-based company Ultrasys and approved by the NHS, and a free online package called MoodGYM, made by the Australian National University Centre for Mental 91ɫƬ Research – provided any additional benefit compared with standard treatment and antidepressants alone. Patients were randomly allocated one of the three options. The researchers checked their progress after 4, 12 and 24 months, which included measuring levels of depression, assessing quality of life, and whether they were back at work.
The results were far less impressive than reported in previous research. Giving people a piece of software and a few supportive phone calls isn’t enough to treat a complex condition like depression, says Gilbody. “People really didn’t engage in the technology in the way we’d expected them to.”
Why the conflicting results? Most other trials have been quite small and some were conducted by people behind the apps being tested, says Gilbody. “There was very little out there in the way of independent evaluation.”
Gilbody concedes that CCBT might provide a stopgap, particularly where there are long waiting lists for face-to-face talking therapies. But he thinks the benefits are slight. The trial also found no difference between the paid-for and free courses. “We think the NHS should stop throwing money at commercially provided products,” he says.
Arguably, the pitfalls of CCBT are somewhat predictable. “It’s like working from home,” says Hillary Rothrock, who used to be a mental-health crisis worker and has herself tried online counselling. “You get out what you put in.” But for many with depression, motivation is a problem. The more support you give people the better they engage with CBT, says Lewis. He thinks the level of support people need to get the clinical benefit from these systems is not yet clear. “If I’m learning French I could go out and buy a CD or I could go to a class,” he says. “The danger with just buying the CD is you can leave it unopened for several months.” Gilbody thinks it’s a good analogy. “I play better guitar than I speak French because I have a guitar teacher turning up every Friday,” he says.
“Using computer-based CBT is like working from home – you get out what you put in”
For Fritz, frequent interactions with a Joyable counsellor were key to her positive experience. He gave her things to read and tasks to do. He was quick to respond to emails when she needed him. “I’ve done CBT before and this was so much better,” she says. Of course, this is anecdotal. But Gilbody’s trial also highlighted the need for interaction – whether for guidance or encouragement. “The thing most participants reported missing was the human contact,” he says.
Other people were also what Jamie Druitt missed when he suffered from depression after the break-up of his marriage. It’s what led him to set up TalkLife, a social network that caters specifically for people with depression, anxiety disorders and self-harming behaviour. “Your mates are just like, let’s go out for a beer, you’ll be fine,” says Druitt, looking back. It wasn’t enough, he felt. “I firmly believed that there were people around the world going through exactly the same thing.”
Of course, he was right. He launched the app in 2014 and within a matter of months it became a place where tens of thousands of people – especially teenagers – were sharing experiences and offering each other support. TalkLife now has 180,000 users from 120 countries. “Kids in Botswana are helping kids in the US,” he says. Druitt’s idea was to take the best bits of social networks like Facebook – the connections, the sharing – and build a safe haven around them. “You can’t put a post up on Facebook or LinkedIn about how you’re actually feeling,” he says. Among the usual online banter, people discuss self-harm, running away, abuse.
Druitt wants TalkLife to be the online therapy that works. Too many people struggle with depression or anxiety without seeking help, he says. He is convinced some just need a place to open up. “We see people come in in a crisis state and then within about 30 days turn around to become amazing helpers, sharing their story with others,” he says.
But what sets TalkLife apart is the work going on behind the scenes. Druitt realised that the millions of posts being produced were an invaluable source of data. So he contacted people he thought could do something with it.
“We noticed immediately that people talk about things on TalkLife they would not talk about in person,” says computer scientist at the Massachusetts Institute of Technology, one of around 20 researchers now working on TalkLife. “We’ve never had data of this kind on this scale before.” The team also includes Matthew Nock at Harvard University, who is a leading researcher on self-harming behaviour, and researchers at Microsoft Research in Redmond, Washington.
As with Facebook and Twitter, Dinakar and his colleagues are using machine learning algorithms to analyse what he calls breadcrumb trails – tracking what users say, who they say it to, what they read, what they like, and so on. “All these behavioural breadcrumbs are used heavily for monetisation on virtually every social network,” says Dinakar. “Most people in machine learning are doing all they can to make people click on ads.” But with TalkLife, the aim is to understand users’ mental well-being – and intervene if they need help.
“TalkLife aims to understand users’ mental well-being and intervene if necessary”
To do this, the researchers have built computer models based on Nock’s detailed characterisation of self-harming behaviours. By applying these to TalkLife data, the team can study such behaviour as it plays out in real life. “Why people engage in self-harm is far more complex and nuanced than clinical models often capture,” says Dinakar. “And they do it in spurts, not all the time.”
at Microsoft Research has previously had success applying a similar technique to postnatal depression. It’s a very under-reported condition, says Counts. “A lot of women don’t even realise they have it. You’re tired, but of course you’re tired, you just had a baby.” By looking at women’s Facebook and Twitter activity before and after giving birth, Counts found he could identify those who suffered some form of depression. Indicators included the number of friends or followers an individual had, whether she asked a lot of questions, and even the words she used. Increased use of the first-person pronoun, which tends to reflect an increased inward focus on the self, is correlated with depression, for example. Counts also looked at whether anyone was talking back. “You could be posting like crazy,” he says. “But are you actually interacting with people?”
He then realised some indicators, such as apparent stress levels, allowed him to make a prediction even before a woman had her baby. “The next step – and this is similar to what’s happening with TalkLife – is to start putting these results into practice,” he says. With postnatal depression, for example, Counts imagines pregnant women using an app that relays information of this kind to their doctor.
A big advantage of using online breadcrumb trails rather than the standard questionnaire to assess mental health is that breadcrumbs don’t lie. “They’re generated in the flow of life and therefore reflect everyday experience,” says Counts. “Pronoun usage, for example, is very hard to fake.” When you ask someone to fill in a questionnaire, they have time to think.
With its growing community of active users, TalkLife now has ample data with which to make its own predictions. Go to a doctor with a cold and she might tell you to take paracetamol and come back if you’re not feeling better, says Dinakar. If you are still ill, she will try something else. “We would like to do something like that with TalkLife,” he says. Druitt hopes to start rolling out automatic interventions next year. Details have yet to be announced, but potential actions include automatically connecting people in trouble with someone who has been through similar experiences and presenting people with CBT-inspired messages designed to help them change their thinking. “You can think of them as our ads,” says Dinakar.
“Social interactions online can be used to predict depression automatically”
For many, the thought of a personal psych profile sitting on a social network’s servers will ring alarm bells. with dodgy data handling practices, for example. The risks of sensitive data being leaked or misused are real. But Druitt says the TalkLife team are open with their users about what they are doing and that the new features will be opt-in only. He thinks the community will be supportive.
In time, such systems could be added to other social networks. Dinakar has been invited to talk at Facebook. And Thomas Insel recently left the US National Institute of Mental 91ɫƬ to head Google Life Sciences. Like the TalkLife team, he thinks machine learning can help us study mental health. It might also improve CCBT.
Ultimately, Counts thinks we might all end up with mental well-being monitors in our pockets. They would track our social lives, crunch psychological metrics and give us regular updates. Fitness trackers are now common, why not mental fitness? “It’s not that different from monitoring your pulse or your glucose level,” he says.
(Images: Dan Page (artwork); Bruno Arbesu/picturetank)
The therapist in the room
“I think when you have a therapist, you have an extra personality in the room,” says Hillary Rothrock, a former mental-health worker based in Maryland. “It influences the therapeutic dynamic.” . “Doing it this way kept it between him and me,” she says. “We weren’t trying to influence anyone else’s opinions.”
If three’s a crowd, two is not necessarily company. A 2014 study by Gale Lucas at the University of Southern California in Playa Vista and colleagues showed that, even in a one-on-one situation, people were more willing to disclose details about their mental well-being to an on-screen avatar if they believed it was controlled by a computer rather than a human. The researchers behind the study suggest that automated “virtual humans” might help therapists gather accurate information from their patients.
Communicating via keyboard can itself have advantages over face-to-face speech, says Glyn Lewis at University College London. “When you talk to people there tends to be a lot of chit-chat,” he says. “By getting people to write things down, you’re getting straight to the point. That may speed up the therapy process.”
This article appeared in print under the headline “Uploading depression”
Article amended on 8 January 2016
Correction: When this article was first published, one of the tested online packages was incorrectly referred to. Thefree, online computer-based cognitive behavioural therapy package tested by Simon Gilbody and his colleagues was actually MoodGYM,made by the Australian National University Centre for Mental 91ɫƬ Research. The hasnow been corrected.
