How AI Can Greatly Improve Mental Health Services

“The future is already here – it's just not very evenly distributed,” ― William Gibson.

Artificial intelligence is creeping into mental healthcare.

“How does it make you feel to hear that?” is the question that you might be already discussing with an AI bot. Even more, your AI therapist might be quite successful at helping you overcome any feelings of worry about a direction our future can take with the rise of AI.

If looking beyond the scary headlines about Skynet coming true, an increasing use of AI in mental healthcare is actually great news for many of us.

Today, we are experiencing a severe mental health crisis aggravated by the COVID-19 pandemic. Called the plague of modernity, mental health disorders affect one-fifths of people in the US, with 40% of Americans developing symptoms of anxiety or depression or both last year due to the fear of illness, self-isolation, grief, or deepening financial concerns.

The landscape around this crisis is no less daunting. Its longtime hallmarks have been the societal stigma around mental illness and an acute shortage of mental health professionals.

AI is already disrupting major industries, from healthcare to education, addressing challenges and opening thrilling opportunities. Mental health may well be the next candidate. Thanks to rapid technological advances, AI solutions for healthcare have all the potential to spearhead a positive change in this space that has been long ripe for innovation.

Mental health stats ― The crisis we are in

Mental health disorders are among the leading causes of morbidity and mortality, expected to cost the world’s economy some $16 trillion by 2030.

They affect at least 10% of the population, with up to 20% of children and adolescents suffering from some type of mental disorder and women more likely than men to be diagnosed with depression.

In the US, around 19% of people experienced a mental illness in 2017-18, which was a year-on-year increase of 1.5 million people. The number of the young reporting mental distress is rising more significantly.

No one seems to know exactly why depression and anxiety are so common nowadays. Many experts even dismiss any upswing, arguing that what we see is a surge of people actively seeking treatment.

Indeed, the number of people applying for help with depression or anxiety in the US has soared, according to the 2021 State of Mental Health in America report. As many as 315,220 people took anxiety tests in January to September 2020, 93% up from the entire 2019. The number of people taking depression screens increased by 62% to 534,784 people.

At the same time, only around 40% of US adults with mental illness received treatment in 2019. One of the reasons is a shortage of mental health providers, which the US National Council for Behavioral Health predicts can reach over 15,000 clinicians in the next couple of years.

The need for solutions that can scale access to mental health treatment is desperate, while such solutions are already here, thanks to AI.

Can AI help with mental health and how?

Unlike medical areas such as radiology or pathology where AI can outperform doctors, mental healthcare is thought by many as an exclusively human field where emotional intelligence is essential. Most mental health practitioners doubt that artificial intelligence solutions for mental health will ever be able to provide emphatic care.

While clinicians may be overwhelmingly skeptical about the use of AI in mental health care, people emotionally connect to technology because they tend to treat robots as living beings, researchers say.

We are not talking here about the unsettling intimate bond developed between a lonely man and an AI operating system in the movie Her, but rather about people’s willingness to pour their hearts out anonymously to an AI bot companion, as found out by creators of Eliza, one of the first chatbots designed in the 1960s.

Moreover, people do prefer to talk to robots. Only 18% would favor humans over technology to discuss their mental health struggles compared to 68% who would choose robots, believing that robots don’t judge, are unbiased, and can provide an instant answer to a health-related question, according to a recent study conducted by Oracle and Workplace Intelligence.

Just as important, talking to technology helps.

Multiple meta-analyses have confirmed that computer-aided cognitive-behavioral therapy (CBT) delivered via desktop or mobile apps is equivalent to or even more effective than standard CBT. The National Institute for Health and Clinical Excellence (NICE) in England first recommended computerized CBT packages for depression, panic, and phobias back in 2006 on the grounds of clinical and cost-effectiveness.” In the US, the COVID-19 pandemic has prompted the Food and Drug Administration (FDA) to relax policies for a broader use of digital therapeutic tools to treat mental conditions.

Particularly important to mental health care are the following AI technologies:

• Machine learning (ML) and deep learning (DL) that provide greater accuracy in diagnosing mental health conditions and predicting patient outcomes

• Natural language processing (NLP) for speech recognition and text analysis that is used for simulating human conversations via chatbot computer programs, as well as for creating and understanding clinical documentation

• Computer vision for imaging data analysis and understanding non-verbal cues, such as facial expression, gestures, eye gaze, or voice tenseness

Examples of how AI is already revolutionizing mental healthcare

Currently, the applications of artificial intelligence in mental healthcare revolve around:

1. Analyzing patient data to assess the probability of developing mental health conditions, classify disorders, and suggest optimum treatment plans.

The data subject to analysis may include electronic health records (alongside blood tests and brain images), questionnaires, voice recordings, and even information sourced from a patient’s social media accounts.

Data scientists employ a variety of techniques, such as supervised machine learning and natural language processing, to parse patient data and flag mental and physical states — i.e., pain, boredom, mind-wandering, stress, or suicidal thoughts — associated with particular mental health conditions.

A group of researchers from IBM and the University of California have recently analyzed 28 studies exploring the use of artificial intelligence in mental health and arrived at a conclusion that, depending on the choice of an AI technique and quality of training data, algorithms manage to detect an array of mental illnesses with 63-92% accuracy.

2. Conducting self-assessment and therapy sessions.

This category is largely represented by keyword-triggered and NLP chatbots that help patients evaluate the progression and severity of a mental illness and cope with its symptoms —either on their own or with the help of a certified psychiatrist waiting on the other end of the virtual line.

An example of this would be Ellie, a digital avatar that was initially designed to help war veterans struggling with depression and PTSD. The AI therapist not only understands words but can also interpret non-verbal signs, such as facial expression, posture, or gestures to comprehend a patient’s emotional state and choose the right words to alleviate stress and anxiety.

AI-powered diagnostic solutions may also work in tandem with wearable devices that measure heart rate, blood pressure, oxygen levels, and other vital signs that indicate changes in a person’s physical and mental well-being. One of such solutions is BioBase, a mental health app that leverages AI to interpret sensor data coming from a wearable. Designed to help companies prevent employee burnout, the mental health tracker reportedly helps reduce the length and number of sick days by up to 31%.

3. Making psychological interventions by automatically giving appropriate information to the patient.

Often deployed as part of a hospital’s digital patient engagement strategy, solutions like these help patients navigate a mental health crisis by providing relevant information and friendly assistance.

An international team of scholars achieved impressive results with Tess, an artificial intelligence chatbot that delivers highly personalized psychotherapy based on CBT and other clinically proven methods, along with psychoeducation and health-related reminders. The interventions are delivered via text message conversation, meaning that emotion identification relies solely on language processing. The researchers tested the chatbot among a group of students to find out that the individuals who conversed with Tess daily over a period of two weeks displayed a significant reduction in mental health symptoms compared to participants who had sessions less frequently. Although Tess cannot possibly fill the shoes of a trained psychiatrist, the chatbot proves to be a viable alternative to human specialists — not the least thanks to its ability to tailor content based on a patient’s diagnosis and demographic data.

4. Equipping therapists with technology to automate workflows, monitor the treatment process, and improve medication adherence.

Due to the very nature of mental health conditions, psychiatrists can seldom rely on legacy tech tools or other physicians’ advice when interpreting medical data and devising treatment plans for patients. One way to lessen the administrative burden could be the implementation of AI-driven mental health platforms that automatically retrieve information from miscellaneous IT systems within a hospital and generate on-demand reports about every single patient’s progress, current condition, and possible outcomes. An early example of such systems is OPTT, an AI platform that provides a rich selection of tools for mental health professionals looking to increase the capacity of their clinic. Preliminary research indicates that OPTT could improve access to quality mental healthcare by up to 400%.

Outside the professional psychiatry realm, there are a plethora of AI-powered mental health apps like Woebot, Replika, and Elomia on the App Store and Google Play that target individuals with mild mental health disorders and extend psychiatric and psychological care beyond hospital doors.

Benefits of using AI in mental health treatment

The current success of artificial intelligence apps and platforms for mental health care can be attributed to the following benefits:

Affordability. Unlike traditional counseling where you need to schedule and travel for appointments, AI-based and other mental health apps allow users to access therapeutic help anywhere, anytime. Moreover, they provide help at little or no cost, compared to in-person therapy rates, missed work, and the need to make other arrangements and commute.

Accessibility. AI-based apps remove such barriers to mental health treatment as staff shortages across the board and a lack of providers in rural and remote areas. This is important since more than 100 million people in the US live in so-called Health Care Professional Shortage Areas. Location-agnostic AI chatbots and platforms can see you whenever you need and spend as much time with you as you need.

Efficiency. Artificial intelligence algorithms used in mental healthcare have already been proven to be successful in detecting symptoms of depression, PTSD, and other conditions by analyzing behavioral signals. Other studies have shown that algorithms make more accurate assumptions than clinicians to distinguish between genuine and fake suicide notes and are 100% accurate at predicting who among at-risk teens are likely to develop psychosis. They also help patients struggling with mental distress: a randomized controlled trial conducted by Woebot researchers has revealed that participants experienced a substantial decrease in depression and anxiety after just two weeks of using the app.

Privacy and ease to open up. AI-based therapists make people feel less self-restrained to share embarrassing things. This is especially important for those who can feel shame in face-to-face interactions because of stigma or fear of being judged. Actually, almost a quarter of people lie to doctors, with the most hushed topics being smoking, drinking habits, and sexual activity. For many, it’s easier to admit the true extent of their behavior to a robot because the robot won’t judge.

Support for therapists. “AI could be an effective way for clinicians to make the best of the time they have with patients,” says Peter Foltz, a research professor at the University of Colorado Boulder. This is because AI can track and analyze substantial amounts of data faster and even more efficiently than any human. As a result, algorithms help with more accurate diagnoses. They can also spot early signs of trouble by monitoring the patient’s mood and behavior and alert clinicians so that they can quickly adjust treatment plans. This can be lifesaving for suicidal patients who need regular check-ins.

AI trends in mental health

According to a recent report, funding in digital mental health increased from $1,136 million year on year to $1,974 million globally in 2020, with mental health seeing an all-time record number of mentions in the news in March 2021.

AI-powered mental healthcare also sees record levels of venture capital funding.

Amid the crisis, researchers have made a few big steps forward in their artificial intelligence projects in mental health that may be a hint of what lies ahead for us in the very near future.

We are likely to see the emergence of more AI therapists like Ellie and some major developments driven by AI prediction and detection capabilities.

For instance, researchers at Vanderbilt University Medical Center in Tennessee, US, have developed an ML algorithm that uses a person’s hospital admission data, including age, gender, and past medical diagnoses, to make an 80% accurate prediction of whether this individual is likely to take their own life. University of Florida researchers are about to test their new AI platform aimed at making accurate diagnosis in patients with early Parkinson’s disease. Another team of researchers is working on an AI tool to predict schizophrenia based on analysis of brain scans. Research is also underway to develop a tool combining explainable AI and deep learning to prescribe personalized treatment plans for children with schizophrenia.

So, it looks like we have the most promising technologies today to overcome the crisis and transform the delivery of mental healthcare.

A final note

AI holds both incredible promises and many next-level complexities, including an array of ethical pitfalls.

Like with many healthcare apps, innovators may need to be compliant with the GDPR, HIPAA, and other industry-specific guidelines. But there’s much more to that with artificial intelligence. Creators of AI mental healthcare technology should be wary AI challenges, including bias that may come with insufficient and poor quality databases, increased risk to patient privacy and data security, a lack of guidance on the development of AI tools, calls for greater transparency over the use of algorithms and their decision-making logic, complexities with integration into clinical practice, and the need to train mental health professionals in emerging technologies.

Scaling access to mental healthcare is critical, but it should be scaled ethically not to lead to increased risk to patients.

If you are interested in learning more about AI opportunities and challenges in mental healthcare or want to innovate in this space, contact our AI experts.