The COVID-19 pandemic took a toll on both the physical and mental well being of humankind. The side effects of the impact have brought in a surge in people seeking help to deal with illness, unemployment and social isolation, all in an increasingly fast-paced, pressure-filled world. However, the digital revolution that has transformed many aspects of our daily lives has not yet truly emerged in behavioural health care. As a supplement to therapy, AI and machine learning now seem to have the potential to revolutionise how we diagnose and treat mental health conditions. Soon, algorithms may become our first line of defence against the mental health struggles that can be debilitating for many.
How can AI aid?
Researchers are now pioneering a new approach to mental health care wherein an AI analyses the language used during the therapy sessions. An automated form of quality control is becoming more and more essential in helping therapists meet the demand. Natural-language processing, also known as NLP, identifies which parts of a conversation between the therapist and client and which types of utterance and exchange seem to be most effective at treating different disorders. Understanding therapy’s most essential ingredients could help open the door to more personalised mental-health care, allowing doctors to tailor psychiatric treatments to specific clients, much needed when prescribing drugs.
Machine learning techniques that carry out the automatic translation from sessions enable quickly analysing vast amounts of the language. This gives researchers access to an endless and untapped data source: the language therapists use. Researchers also believe that they can use insights from processed data to boost therapy and its outcomes. This can result in more people getting better and healing fast.
Sign up for your weekly dose of what's up in emerging technology.
At first, a few hundred transcripts are annotated by hand to train the NLP models, highlighting the role therapists’ and clients’ words play at that point in the session. The technology works similar to a sentiment-analysis algorithm that can distinguish and tell whether movie reviews are positive or negative. Next, the AI translates natural language into a kind of bar code or fingerprint of a therapy session that reveals the complete role played by different utterances. For example, a fingerprint for a session can show how much time was spent in constructive therapy versus general conversations. Such a readout can help therapists focus more on the former in constructing future sessions. AI techniques could also help prospective clients match with therapists and determine which types of therapy will work best for them.
A virtual therapist named Ellie was also launched and trialled by the University of Southern California’s Institute for Creative Technologies (ICT). Ellie was initially designed to treat veterans experiencing depression and post-traumatic stress syndrome. It can detect words and nonverbal cues such as facial expressions, gestures, and different postures. Nonverbal signs are an important aspect in therapy yet can be subtle and difficult to pick up. Ellie’s creators argued that this virtual human can advance mental health and improve diagnostic precision.
A few psychologists argue that humans might find it easier to share potentially embarrassing information with a virtual therapist, whereas, in human-to-human interaction, there often seems to be a degree of self-restraint. It has been observed that when patients talk to a therapy bot, they report not feeling judged. Therapists, on the other hand, can discuss the AI-generated feedback for further improvements. The idea is to help therapists take control of their professional development, showing them what they’re good at, things that other therapists can learn from, and a few of them not so good at things they might want to work on.
Although AI for mental health still needs to deal with many complexities, research shows that behavioural health interventions benefit from continuity, and technology seems to offer an improved user experience. Furthermore, while the human brain is complex with its own set of challenges, data collection from behavioural health sessions in a consistent, measurable and accessible manner will be essential to better care and better results in the near future.