Using artificial intelligence (AI) to improve CBT and positive mental health
In the U.K, the national institute for health and care excellence (NICE) recommends cognitive behavioural therapy (CBT) as a treatment for common mental health problems. Talking therapies such as CBT are widely used as low-intensity interventions for depression, generalised anxiety disorders, panic disorders, obsessive-compulsive disorders, post-traumatic stress disorders and social anxiety.
How does CBT work for anxiety related disorders
Cognitive behavioural therapy has long been identified as one of the most effective ways to cope with stress, everyday challenges, complicated relationships and help users deal with grief etc. It is based on a psychological model typically delivered over 10 to 20 weekly sessions. The talking therapy is centred on helping users:
- Recognise anxiety: identify feelings, bodily or somatic reactions to anxiety
- Clarify thoughts: identify and understand anxiety-provoking situations. Clarify any thoughts and cognitions in these situations.
- Develop coping skills: In lieu of anxious talk, thoughts and actions, users can develop unique coping self-talk and thoughts for coping purposes.
- Evaluate outcomes: crucial step in helping users develop a beneficial way of thinking and behaving. The purpose of CBT is to create a unique understanding of harmful feelings, thoughts, behaviours etc. Use this information to develop coping/behavioural strategies and ultimately reduce the occurrence, intensity etc. leading to an improved quality of life.
How is artificial intelligence improving CBT?
The Merriam-Webster dictionary defines artificial intelligence as the simulation of intelligent behaviours in computers. It encompasses any capabilities for a machine to imitate intelligent human behaviours and make decisions based on logic.
Artificial intelligence (AI) has undoubtedly created groundbreaking opportunities for a wide range of industries ranging from healthcare, to finance and transportation. In reference to Roger’s innovation curve, the adoption of AI technology in the healthcare industry is predominantly driven by the “early adopters” and the “early majority”. The increased adoption of artificial intelligence and it’s advancements are creating new opportunities for the millions of people affected by common mental health conditions.
Machine learning plays a crucial role in improving the delivery of CBT. In the simplest of terms, the machine’s decision making process is created by training the machine learning with data. Post-training, when a user provides the model with an input, a logical decision/output is provided by the machine.
Large sets of data are often required to train the machine learning model to a stage it can actively be used in a real-time environment. Phobot, a voice and text operated conversational agent uses verbal and written cues to help assess the user’s anxiety. The virtual therapist then uses its training to deliver personalised, purpose-built CBT programmes to treat anxiety. Dr.Simon Lewis has worked with Cadscan in the development of Phobot to ensure “interactions with the user are as close to real life therapy sessions as possible. We’ve done this by adhering to widely used CBT models of understanding and working closely with social anxiety as well as running back to mock sessions which are then transposed into adaptable scripts for Phobot”.
Though complex, the process could undoubtedly increase accessibility to mental health services, particularly for the most vulnerable and difficult to reach populations. Dr Alastair Buchanan, Managing Director of Cadscan, said: “Many young people find it hard to access treatment for anxiety. They can develop poor coping mechanisms, including avoidant behaviours or alcohol and drug use, which lead to further problems and limit their life chances. Phobot unobtrusively delivers effective, evidence-based support whenever they need it.”
By continuously developing it’s experience, machine learning is incrementally improving the delivery of CBT. It’s experience is captured at a fraction of the cost, time and labour required in a conventional therapeutic environment (days compared to years). Moreover, the experience built is used to refine and personalise treatment for many individuals at a mass scale at a fraction of the cost.
A synergy of natural language processing and machine learning has been used to engineer Phobot. The virtual therapist is a true chatbot capable of holding basic conversations with a user, determining the context of the conversation and only delivering CBT when required. When anxiety is detected, Phobot will ask about the user’s mood and thoughts, the virtual therapist will begin to listen to how the user is feeling and deliver personalised, evidence-based CBT. Machine learning is then used to evaluate outcomes, and could potentially predict the likelihood of anxiety provoking situations (based on data).
Phobot will be available to smartphone users to create an easily-accessible, just-in-time, personalised approach to CBT. The pocket-sized virtual therapist will be available 24/7 and features a low friction, dialogue-based interface to facilitate engagement and adherence.
For more information, kindly enquire at
- National institute for health and care excellence: Common mental health problems: identification and pathways to care. CG123. https://www.nice.org.uk/guidance/cg123/ifp/chapter/what-treatments-might-i-be-offered
- James AACJ, Soler A, Weatherall RRW, Cognitive behavioural therapy for anxiety disorders in children and adolescents (Review). 2005. https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD004690.pub2/epdf/full
- Merriam Webster, Artificial intelligence. https://www.merriam-webster.com/dictionary/artificial%20intelligence