The Transformative Potential of AI in Reshaping Mental Health Care by Dr Ashkan Eliasy
In this article, I examine the profound impact of AI on mental health care within community settings, as AI continues to advance across various industries. My exploration will encompass both current and prospective applications of AI in this field. Furthermore, I will delve into the realm of AI-powered apps that enhance the accessibility of therapies like CBT, while also considering the risks associated with an unregulated industry. Ultimately, the article will provide insights into the potential future landscape of mental health care, with AI playing a pivotal role.
Artificial Intelligence (AI) and mental health care are joining hands. This combination promises significant changes in how we tackle mental health challenges, helping us to live healthier, more fulfilled lives. It’s about understanding and treating mental disorders, a big issue worldwide, affecting over a billion people. Since 2017, over $10B has been invested in the digital mental health industry. This has led to 10 companies reaching a valuation of over $1B and millions of patients receiving care from companies that didn’t exist five years ago.
Mental disorders make life challenging. There’s no one-size-fits-all way to diagnose them, making their treatments complicated. Doctors use guidelines like the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) and the International Classification of Diseases (ICD-11) and listen to patients’ stories to figure out what’s wrong. This takes time, effort and resources. Here’s where AI can make a difference. Imagine using AI to analyse piles of data and find patterns. Patterns can help diagnose mental disorders, making the process faster and more accurate. We’re already seeing AI being used innovatively, like social bots for dementia care and virtual psychotherapists. These tools can spot issues early and personalise treatments, helping people feel better sooner.
AI comes into play in three main ways. Firstly, it helps refine therapeutic interventions, as seen in platforms like ieso and Lyssn. They use AI to analyse therapy sessions and offer feedback to therapists. Secondly, AI can detect signs of mental health issues in medical records, like the REACH VET program does for US military veterans. Lastly, AI may guide treatment in fully digital settings. Platforms like Woebot, Wysa, and Tess have integrated AI into therapeutic chatbots for years, but there’s still debate about their effectiveness.
Chatbots have been part of the mental health landscape for 60 years, working towards making therapy more efficient. Today, with fewer than 4 psychiatrists per 100,000 people globally, these digital helpers are more important than ever. They’re built into thousands of smartphone apps, aiming to make therapy accessible for everyone. However, it’s crucial to question how well these apps work. With proper scientific evidence, we can be sure if they’re helping or just giving false hope.
Cognitive Behavioural Therapy (CBT), a form of psychological treatment, is a crucial area where AI can play a significant role. CBT helps patients understand and change thought patterns leading to harmful behaviours or distressing feelings. With AI, it’s now possible to create digital platforms that deliver CBT in interactive, engaging ways. These platforms use AI algorithms to adapt treatment plans based on a user’s input, making therapy more personalised and accessible. Whether it’s managing anxiety, stress, or depression, these AI-powered CBT tools can provide support anytime, anywhere, breaking down traditional barriers to costly and often inefficient mental health treatment.
Another promising application of AI is in exposure therapy. By simulating real-world experiences in a controlled virtual environment, AI can help patients tackle their fears more effectively. For example, seeing an AI-generated younger version of oneself can boost mental health and well-being. AI also helps us hold onto memories of loved ones who’ve passed away. These digital remembrances can serve as ‘technological heirlooms’, providing a new means for reflection and remembrance. From recreating a deceased friend as a chatbot to animating old family photos, AI offers new ways to cope with grief and loss.
People suffering from facial deformities or speech disabilities due to disorders like amyotrophic lateral sclerosis (ALS) can benefit from AI too. They can regain their voice and facial expressiveness in digital interactions, giving them a sense of normalcy. This has been a game-changer for many individuals who’ve lost their ability to communicate naturally.
Despite the promise and potential, the reality is that AI in mental health care is still very new. There’s a lot we don’t know and a lot that could go wrong. Predictive models often fail when applied across different populations and institutions. Cultural, legal, and structural challenges can limit their adoption. There’s also the issue of trust – can we really trust an AI with our mental health?
The app market is flooded with mental health applications, with estimates showing that more than 10,000 mental health-related apps are available worldwide. While this indicates a robust effort to leverage AI in mental health, it also brings to light a considerable challenge. According to a study published by the Journal of Medical Internet Research, around 3% of these apps had been validated through peer-reviewed research, suggesting a vast majority lack scientific evidence supporting their effectiveness. Additionally, a study in Nature Digital Medicine found that user retention rates for such apps are less than optimal. Roughly 28 days after downloading a mental health app, only about 4% of users were still engaged with it. This significant drop-off points to an apparent inconsistency between the promise of AI-assisted mental health care and the user experience. These issues raise important questions about the quality and reliability of these digital tools in managing mental health effectively.
With an industry on the rise, it’s essential to question its ethical side. What happens when users stop using these apps after a few weeks? Is it ethical to push them to keep using the apps? Privacy is another major concern. Several incidents have raised eyebrows about how these platforms handle sensitive user data. With the demand for data to train AI, privacy issues will continue to be a pressing concern. A recent report published by Mozilla Foundation found insufficient privacy policies in 70% of mental health apps. This means with digital therapy; personal details can be at risk.
Systems like HAILEY showcase an intriguing direction for human-AI collaboration in mental health care. HAILEY is an AI-in-the-loop agent designed to enhance empathic conversations during peer-to-peer mental health support. It operates by providing real-time feedback to peer supporters, assisting them in responding more empathically to those seeking help. HAILEY showed potential in a trial with real-world peer supporters on TalkLife, a prominent online peer-to-peer support platform. This human-AI teamwork approach led to a 19.6% boost in empathic conversations between peers overall. More strikingly, there was a 38.9% surge in empathy among peer supporters who found it tough to provide support. Analysing these human-AI interactions showed that peer supporters used AI feedback well without becoming too reliant on it. They also reported feeling more capable after receiving feedback.
Studies are pouring in on AI’s ability to diagnose mental health issues like Alzheimer’s, Schizophrenia, bipolar disorders, and post-traumatic stress disorder. AI is showing some promising signs. For instance, it can look at brain images and tell if someone has Alzheimer’s. It can even spot patterns in people’s talk and predict if they may get a mental disorder. The peer-reviewed data suggest up-to-perfect performance and a high likelihood of clinical adaption for accelerated, more accurate and objective diagnoses of mental health disorders. But let’s not get too excited yet. These are early days, and researchers have a lot of work to do to ensure these AI tools work well, are fair to everyone, and respect our privacy.
In addition, laws need to keep up, too. For example, who’s to blame if an app gets a diagnosis wrong? The app maker? The AI coder? Or the data the AI learned from? Maybe there should be a human double-checking the AI’s work. Or clear rules to hold someone accountable. As we fine-tune these tools, we must remember to make them fair, transparent, and safe for everyone. AI in mental health care has great potential. It can help more people in better ways. But we must stay sharp to avoid any pitfalls along the way.
By Dr Ashkan Eliasy
Featured Photo by Barbara Zandoval on Unsplash
