You are here

An intelligent approach to mental health

May 27,2019 - Last updated at May 27,2019

BOSTON — A few years ago, toward the end of his life, my father battled severe depression. As a physician and professor, he did not lack access to mental healthcare. But he had grown up in a society that stigmatised mental illness, and he was unwilling to seek professional help. As a son, it was devastating to watch my father suffer. As a public-health researcher, I gained a new awareness of the myriad systemic failures in the provision of care.

Scientists from around the world are now seeking to address the problems with “Countdown Global Mental Health 2030”, a “multi-stakeholder monitoring and accountability collaboration for mental health” launched in February. But, while this initiative is a positive step, it neglects a key element of an effective solution: advanced technology, especially artificial intelligence (AI).

Globally, the supply of psychiatrists and clinical psychologists is nowhere near sufficient. For example, in Zimbabwe, there are just 25 mental-health professionals for a population of over 16 million. While the country has produced some innovative and useful community-led initiatives, such as the “Friendship Bench,” their scalability is limited.

Lack of access to mental healthcare is hardly a developing-country problem. In the United States, almost half of the population is unable to access comprehensive mental healthcare, often owing to financial constraints.

Beyond access, there is the stigma issue, exemplified by my father’s experience. Clinical evidence indicates that stigma takes two forms. People who seek mental healthcare may face public stigma in the form of discrimination and exclusion, owing to endemic misconceptions about mental illness. When those beliefs are internalised, sufferers may also struggle with self-stigma: low self-esteem, low self-efficacy and unwillingness to pursue productive opportunities.

The consequences of failing to provide adequate care have been severely underestimated. According to one study, mental-health issues are responsible for 32.4 per cent of years lived with disability and 13 per cent of disability-adjusted life years, years of “healthy” life lost due to disease, disability or untimely death.

The economic costs are enormous. According to a 2015 analysis, in the US alone, the total economic burden from mental health exceeds $210 billion annually. More than half of that is attributed to workplace absenteeism and productivity losses; another 5 per cent is attributed to suicide-related costs. Companies’ efforts to circumvent the need for mental healthcare by reminding workers to practice mindfulness are probably not as helpful as proponents claim.

What could help are AI-based solutions, such as chatbots. By mimicking natural language to sustain a conversation with a human user, these software systems could act as virtual therapists, providing guidance and support to those who have no alternatives. A randomised control trial reported by clinical psychologists from Stanford University showed that chatbots were significantly better at reducing the symptoms of depression than an information-only approach.

The sort of provisional mental healthcare provided by chatbots would be particularly useful in communities with an inadequate supply of trained professionals. At a time of unprecedented access to smartphones in developing economies, Internet-based solutions would amount to a boon for mental-health accessibility.

Chatbots could also help overcome the stigma problem, because they can engage people who are otherwise reluctant to seek mental healthcare. A recent study found that up to 70 per cent of patients are interested in using mobile applications to self-monitor and self-manage their mental health. Once people initiate contact with a chatbot, another study indicates, they tend to express themselves more freely than they would with a human therapist, underscoring the priority people place on maintaining privacy and avoiding judgment when seeking to address a mental-health issue.

It is now up to clinicians, such as psychologists, to collaborate more extensively with AI developers. Several US universities have already launched programmes that connect experts from clinical sciences with software engineers. These partnerships should be expanded to include universities, especially in countries with a large unmet need for mental healthcare, in order to support the development of linguistically and culturally appropriate virtual therapists.

Involving more diverse actors in the development of algorithms would also help to address the issue of racial and gender discrimination that has cropped up in AI research. Researchers should use fully representative test groups, while taking care to adhere to stringent privacy and accountability protocols.

Of course, such initiatives cost money. Venture capital companies now spend $3.2 billion annually on global health research and development. They should expand the scope of their investments to include AI-enabled technologies for mental healthcare delivery. They could also fund competitions among socially conscious technology entrepreneurs, in order to spur further innovation in this area.

To be sure, AI-enabled mental-health interventions would not, and should not, replace human psychologists or psychiatrists. A chatbot cannot, after all, project real empathy. What it can do is screen for high-risk individuals, such as those with suicidal ideation, and potentially avert destructive behaviour in the short term.

Demand and need often drive innovation. Unfortunately, that has not been true of mental healthcare. It is time to invest in long-term, cost-effective and scalable solutions that build mental healthcare capacity. That effort must include expanded support for traditional services. But it should also take advantage of cutting-edge technologies like AI.


Junaid Nabi is a public-health researcher at Brigham and Women’s Hospital and Harvard Medical School, Boston. Copyright: Project Syndicate, 2019.

27 users have voted, including you.

Add new comment

This question is for testing whether or not you are a human visitor and to prevent automated spam submissions.
4 + 5 =
Solve this simple math problem and enter the result. E.g. for 1+3, enter 4.


Get top stories and blog posts emailed to you each day.