Skip to content Skip to footer


Is artificial intelligence the future of mental health care?

by Victoria Richards, Content Specialist, 26 November 2019

Read Time: 5 minutes

‘Artificial intelligence’ (AI) can conjure up all kinds of sci-fi imagery, often involving cold, calculated robots. In reality, AI simply denotes a machine’s ability to interpret and learn from vast amounts of data with amazing speed, helping us to identify patterns. In fact, beyond its dystopian depictions, AI could actually help us solve one very significant problem: our own mental health.

The mental health crisis

In the UK alone, one in four adults will experience a mental illness at some point. Suicide is the biggest killer of men under 45 and the average wait for effective treatment is 10 years. Factoring in treatment, social support costs, lower employment and productivity, poor mental health is also thought to cost the UK around £94bn a year.

Whether you consider it from an individual standpoint, or the effect on the entire country, we are in the midst of a mental health crisis. Over the last few years, charities such as Mind, Heads Together, CALM and many more have done a wonderful job of reducing the stigma around mental health – but there is still a lot of work to be done when it comes to actually treating people.

That’s where AI can help.

The art of conversation

With waiting times to receive help for mental health issues often taking months, there has been an upswing in companies offering alternative solutions for mental health care.

Among the more well-known examples is Woebot, a ‘charming robot friend who is ready to listen, 24/7’. Created by Clinical Psychologists and experts in AI, Woebot is paving the way for accessible mental health care. Using tried-and-tested methods from cognitive behavioural therapy (CBT), Woebot checks in with users each day and runs through an exercise, such as practising gratitude, or understanding the significance of the language they use to describe themselves.

With each day, the chatbot builds up an understanding of a user’s mood over time, and provides over 150 exercises and lessons to help users reflect in a healthy, productive way. There are plenty of opportunities to provide feedback, ultimately helping Woebot learn more about how to engage with an individual and what resonates with them. Perhaps most promisingly, the creators of Woebot have evidence to suggest their chatbot can have a positive impact in only a matter of weeks. The same can be said of other similar apps, like Wysa and Flow.

A step further on from a messaging interface is the concept of a virtual therapist. This is rather more involved than selecting inputs with a chatbot, and actually involves sitting in front of your computer and talking to an AI assistant.

Ellie, created by the University of Southern California’s Institute for Creative Technologies (ICT), has been ‘trained’ to assess veterans returning home from duty. Incredibly, the technology can pick up on non-verbal cues, such as facial expressions and posture, which can go a long way to helping to analyse a user for symptoms of PTSD.

This type of interface offers a variety of advantages for users. Given Ellie isn’t a human therapist, it can be available anytime, whenever it’s needed. Ellie’s virtual human avatar is also thought to help users ‘open up’ more. The human appearance provides the familiarity needed to strike a rapport, but the AI aspect helps maintain anonymity and prevent fear or judgement that some might feel talking to a human therapist.

All of which presents a pretty fascinating case for an increasingly AI-based talk therapy system. But talking is one thing – how can AI help with mental health in other ways?

Photo credit: Teresa Dey/USC Institute for Creative Technologies

Towards a faster diagnosis

According to Bipolar UK, it takes 10.5 years to receive a correct diagnosis for bipolar in the UK – and before it’s diagnosed, there is a misdiagnosis an average of 3.5 times.

Around 1 in 100 people will be diagnosed with bipolar and in the UK, it’s the fourth most common mental health problem, after anxiety, depression and schizophrenia. It goes without saying that even being able to get a correct, faster diagnosis would change many people’s lives for the better. This is where AI could have a major impact.

Researchers in the US are working to train AI to diagnose mental health conditions, based on patterns they find in everything from surveys and interviews, to MRIs and blood samples. Powerful computers are crunching through this vast amount of data – which would be nigh on impossible for a human – to spot the diagnostic signal. There’s a fascinating, in-depth write-up about this work on The Verge, which is well worth reading through to appreciate just how complex and outstanding this process really is.

Of course, just as there are big positives if all this research goes well, there are risks too. With humans defining the algorithms, there’s a chance that unconscious biases could make their way into the diagnostic system. This needs particularly careful monitoring to ensure people can get the right help – whoever they are.

A brighter future

While we may be in a place where the public simultaneously fears and celebrates technological developments, advances in AI do seem to be moving us towards affordable, readily available mental health care. Indeed, it was suggested at Microsoft’s Future Decoded event that for all the ethical implications people may worry about, it could be argued that not using AI to offer mental health care would be unethical, given how cheaply it can be done and the positive effects it has already demonstrated.

In any case, the success – and continuous development – of apps, bots and systems related to mental health suggests that the public is increasingly keen to see more. With these technologies already learning how to interact with not just data, but people themselves, the future looks brighter for those of us navigating daily life with mental health issues.


Insights straight to your inbox

Sign up now to be the first to know when our experts publish a new article.

Your information helps us monitor and improve our content, we will never share it with anyone else. We only send emails we think you’ll find valuable – you won’t hear from us more than once a week. You can unsubscribe any time.