Articles/Transforming mental health care with AI-powered telehealth

Transforming mental health care with AI-powered telehealth

The COVID-19 pandemic saw telehealth consultations become the new normal in healthcare service delivery. Now the combination of telehealth and artificial intelligence (AI), including AI-powered chatbots, is set to transform the delivery of mental health services globally.

AI-powered telehealth is quickly disrupting the way mental health care is delivered — gathering information to facilitate diagnosis, treatment, prevention and support, especially for individuals who feel uncomfortable talking to a human about their problems.

AI: a helping hand for those in need

In a new telehealth world, AI can mimic the in-person patient–clinician experience. Conversational, AI-based remote monitoring and diagnostic tools allow patients and physicians to interact in a way that closely reflects an in-person experience, no matter where they are located.

AI has the ability to analyse and screen for mental health conditions remotely and at scale. By reading facial cues and analysing voice patterns during telehealth consultations, AI can identify distress indicators and monitor for signs of increasing anxiety, agitation and psychiatric distress. It can then provide feedback and alert care teams if it detects a worrying change or a deterioration in a patient’s mental health.

According to Mia Lander, a psychologist presently working with KPMG, diagnosing and identifying changes early is a crucial benefit of AI in mental health care.

“Where AI can be extremely effective is in the diagnostic area and looking at people’s agitation levels. We have years of research in the psych world looking at distress indicators that would be invaluable for telehealth,” she said.

“AI won’t be telling you this person has schizophrenia — you probably already know that — but saying ‘Hey, this person’s on the upper or down phase of a manic episode’ would give you another valuable piece of information as a professional.”

Putting puzzle pieces together


Human-like AI-based avatars or digital people will execute many of the care tasks human clinicians currently undertake. That includes remotely leading patients through typical questions that clinicians and nurses would usually ask about symptoms and mental health histories before or during consultations.

How does a digital human do that? The AI or digital person can use audio and video functionality to assess language and body language, and to observe a patient’s general mental health condition. By doing that, it can pick up concerns or indications of psychological distress. Those findings can then feed back to the human professionals to commence treatment or fast-track action.

Helping patients who prefer talking to bots

While mental illness rates skyrocketed during the pandemic, the stigma around mental health means many people avoid talking about their problems and seeking treatment. Others feel uncomfortable discussing sensitive topics such as trauma, substance use, sexual history and self-harm. This is a problem because successful therapy depends on a patient disclosing and discussing their history and issues.

I recently spoke about this very issue with Michael Spiegel, the General Manager of State Trustees. Michael explained that young people, in particular, were reluctant to talk and open up, sharing more information when they were not engaging in face-to-face communication with another person, especially with adults.

“When I counselled at-risk teens, I drove them around in my car. That’s when they would open up and talk because they didn’t have to make eye contact.”

How do you get people to discuss their issues when they’re reluctant to open up to another person, including their clinician? By introducing an AI bot instead.

Researchers have found that some users are more open when reporting mental health symptoms to a conversational AI bot than to a human listener. A study focused on building mental resilience among young people showed that users who talked with an AI chatbot had a 45% reduction in depression and a 10-fold increase in taking their medication as prescribed.

Studies have also shown that those patients who prefer speaking with a bot chose to have most of their conversations — which usually included taboo topics — between 10 pm and 3 am, outside typical consulting hours for a human mental health specialist.

Lower costs, faster and better access to treatment

In an already-stretched mental healthcare system, AI has the potential to reduce costs and increase access to treatment.

By not relying solely on the availability of human clinicians, AI-powered telehealth can reduce the cost and societal burden of delivery and treatment of mental health issues. More people get seen, and at a lower cost.

In a world where demand for mental health services outstrips supply, AI can also mean greater continuity of care. Researchers found patients could have a longer-term relationship with conversational AI than with a human clinician, particularly those who move between clinics and practices, such as in locum roles. This is important in regional areas of Australia where clinicians often work in locations on six-month contracts —a massive impediment to continuity of care.

Another benefit of AI is shorter wait times. By augmenting specialist clinicians with AI in telehealth, cases can be addressed and progressed while reducing wait times for an appointment and pressure on already-stretched mental health services.

AI has the ability to make mental health treatment more accessible. Mental health practitioners can use AI to support and monitor patients in lockdown/quarantine, as well as those who have mobility issues, are reluctant to leave their homes, or are in rural and remote areas.

Augmenting, not replacing


AI will increasingly play an essential role in mental health telehealth delivery, but it won’t be at the expense of human practitioners.

A research study published in Frontiers in Psychiatry considered four models for incorporating conversational AI in psychotherapy: clinician only, AI only, and two blends of both — ‘human delivered, AI informed’ and ‘AI delivered, human supervised’. The results suggested that a blended approach would be more suitable than human or AI only.

I have already been a strong advocate for supporting those with mental health issues, and I agree that conversational AI won’t replace human therapists — it will augment the human element to optimise results.