An op:ed written by Stephan Rabie, John Joska, Morgan Watson on the future of AI in mental health treatment in South Africa.

To view the published article on the Daily Maverick's page vist, https://www.dailymaverick.co.za/article/2023-02-20-mind-the-technological-gap-will-ai-close-or-widen-the-mental-health-treatment-gap/

 

Beyond the therapy room, artificial intelligence holds the potential to address mental health on a large scale. There has been an exponential rise in AI-powered, cognitive behavioural therapy-informed chatbots, such as Woebot and Wysa.

Technological advances over the past two decades have revolutionised healthcare. We can reproduce internal organs and bones through 3D printing, consult experts on the other end of the world via advanced communication technology, and monitor our caloric intake and sleep cycles through mobile apps.

Due to these advances, physical illnesses have traditionally been more likely to receive treatment compared with mental disorders, both locally and globally.

However, the latest breakthroughs in the field of artificial intelligence (AI) promise significant improvements in mental healthcare. Our social interactions and news feeds are dominated by debates about the applications of ChatGPT, using Woebot as your therapist, or your local art gallery displaying human-DALL-E collabs.

While several prominent sectors of society are ready to fully embrace the potential of AI, what will it mean for the field of mental health? Will AI replace current mental health practitioners entirely? Or just enough to fill the tragic and increasing mental health gap?

Despite apparent concerns, AI applications in the mental health field are steadily increasing.

Mental illness manifests uniquely in each individual and is best explained holistically by our biopsychosocial profile. The pathophysiology of mental illness is complex and varied, and our current understanding of the impact of biological, social, and psychological interactions is limited.

However, there is evidence that AI can be used to develop advanced diagnostic screening tools and formulate risk models to determine an individual’s risk of developing mental illness. Ultimately, AI and deep learning can be used to identify which treatment techniques work best for which combination of symptoms.

Moreover, AI-powered algorithms can filter through medical histories, family histories, previous treatment plans, and patient behaviour to make informed decisions about appropriate treatment and choice of mental healthcare professionals. In traditional practice, this can be a time-consuming and costly process — the use of AI promises efficiency both in terms of time and money.

Beyond the therapy room, AI holds the potential to address mental health on a large scale. There has been an exponential rise in AI-powered, cognitive behavioural therapy (CBT)-informed chatbots, such as Woebot and Wysa.

Rather than relying on face-to-face therapy, these mental health chatbots are evidence-based and offer users a convenient and discreet way to check in and address mental health concerns. These chatbots ask questions like, “What’s bothering you?” or “How have you been doing?”.

Based on user responses, the apps deliver supportive messages or advice about managing the presenting concerns. These responses are generated from a database that has been prewritten by a psychologist trained in CBT.

In the US, Woebot has been found effective in addressing depression, anxiety, and problematic substance use in university students. For individuals who experience stigma related to their mental health issues, chatbots may be preferred and more accessible compared with in-person therapy. 

A future where mental illness is better managed sounds great — especially in today’s climate where most people with mental illness do not receive any treatment. In South Africa, only one in 10 people living with a mental illness receive the care they need. The demand for better mental health services has increased.

More importantly, meeting these demands has become increasingly difficult and costly due to a lack of resources. For instance, in the South African public sector, there are 0.97 psychologists and 0.31 psychiatrists per 100,000 people.

Can AI address the shortage of mental health professionals? More importantly, will AI replace “human” therapists, psychologists, and psychiatrists? 

The relationship between a therapist and a client — the so-called therapeutic alliance — is considered an important outcome of psychological therapy. A healthy therapeutic alliance relies on relational chemistry between the therapist and client and is supported by human response to verbal communication, non-verbal cues such as facial expressions, and empathy — all human elements needed to drive recovery in some people.

In-person therapy is enhanced by incorporating at least four human senses (vision, auditory, touch, and olfactory), whereas AI is restricted to the first two senses.

Although there is some evidence that people establish emotional bonds with AI mental health chatbots, people often enter therapy vulnerable and suffering, and it remains unclear whether AI therapy is beneficial in this population, especially when compared with therapy provided by human therapists.

Before we concern ourselves with robots taking over, there is one significant barrier to the application and utilisation of AI: access to technology.

Access to computers, the internet, smart devices, and tablets is not ubiquitous in all parts of our country. In impoverished communities across South Africa, access to technology is unreliable.

Our research in peri-urban communities in and around Cape Town found that smartphones, if available, are often shared in households. In addition, even if someone has access to a smartphone, access to free Wi-Fi or mobile data is a common challenge.

Considering that smart devices are often shared, the promise offered by AI-powered mental health chatbots is compromised by potential breaches of confidentiality. We know that poverty is a key social driver of mental illness, and the material deprivation associated with poverty may make AI inaccessible to those that need alternatives to mental healthcare the most.

This does not mean that AI does not have a place in mental healthcare in South Africa and other resource-limited settings. In the spirit of innovation, we need to be creative in promoting access to AI-informed mental healthcare.

Are communal computers in primary care clinics, linked to an AI mental health chatbot, the potential solution? Or are free Wi-Fi hotspots in communities where individuals can access teletherapy the answer?

What we do know is that AI, in its current format, will leave those without reliable access to the technology behind, and only widen the mental health treatment gap. 

Dr Stephan Rabie is a research psychologist and Senior Research Officer in the HIV Mental Health Research Unit, Department of Psychiatry and Mental Health at the University of Cape Town. 

Professor John Joska is Head of Clinical Services (psychiatry) at Groote Schuur Hospital and Director of the University of Cape Town HIV Mental Health Research Unit. 

Morgan Watson is an MPhil in Behavioural Medicine candidate and Content Developer in the HIV Mental Health Research Unit, Department of Psychiatry and Mental Health, University of Cape Town.