Linguistics Analysis for Mental Healthcare
Improving Mental Healthcare through a Tool for Effective Clinical Communication
Mental illness is the biggest single cause of misery in UK society, affecting one in four people and with an economic and social cost estimated at £105 billion. Communication is a critical issue for mental health care, and treatment outcomes depend on the quality of communications they receive from their doctor.
Every 36 hours, the NHS deals with over a million patients, generating multiple spoken interactions between patients and care teams.
The quality of these interactions can have a huge impact on treatment outcomes; in mental health care, communication failure is both common and frequently safety critical.
Researchers from Queen Mary's Centre for Multimodal AI, including Professor Pat Healy and Dr Matthew Purver, set out to analyse and improve communication in a clinical setting. Working in collaboration with colleagues at the University of Exeter's Medical School and the University of Warwick, their research focuses on the basic mechanisms that underpin human communication, and the processes through which people detect and recover from miscommunication.
Improving the quality of health communication
DIALOG+ is a mental health intervention that was conceived by Professor Stefan Priebe in the Unit for Community and Social Psychiatry. The DIALOG+ application was built in collaboration with Professor Pat Healey in the School of Electronic Engineering and Computer Science.
The application is designed to structure conversation in patient-clinician meetings around a series of questions designed to elicit patients' views on their quality of life and treatment. Randomised control trials show that DIALOG+ improves quality of life, reduces symptoms, and achieves a better outcome for independent living, work and social relationships.
The development continues with a new remote version of Dialog+ currently being built in collaboration with Professor Vicky Bird in the Wolfson Institute of Population Health.
DIALOG+ in the real world
The DIALOG+ app is distributed by the East London NHS Foundation Trust. Since 2016 the DIALOG+ intervention has been recommended for use in the mandatory evaluation of all adult patients in East London NHS Foundation Trust and an increasing number of other NHS services benefitting several thousand patients in England and Wales every day.
The DIALOG+ App has been translated into over 15 languages and has been implemented and tested in various studies in more than 10 countries (including Austria, Argentina, Bosnia and Herzegovina, North Macedonia, Serbia, Kosovo, Montenegro, Peru, Colombia, Uganda, India, Pakistan) on four continents.
Providing the patient perspective
As part of their research, the team studied the effectiveness of particular conversational strategies in clinical interactions. Using their findings, they created a communication training programme for clinicians treating psychosis, known as TEMPO. TEMPO helps clinicians understand how it feels to experience psychosis by using simulations of 'hearing voices'.
Dr Michelle Gilmore, consultant psychiatrist at Sunshine Coast University Hospital, Australia, reported that TEMPO helped psychiatrists and medical students understand how to assess and build rapport with patients presenting with psychosis. She said: "This is crucial for engaging these patients in treatment, many of whom do not consider themselves to be unwell or in need of treatment. When doctors can engage these patients in treatment, they are much less likely to relapse and end up back in hospital."
Clinicians also reported a change in their perspective on how patients feel the need to talk about and make sense of their experience. As a result of the training, doctors became around six times more engaged with their patients.
Based on this success, TEMPO is now being delivered by 23 NHS Trusts nationally and 41 institutions internationally, with more than 600 clinicians treating over 550 patients each a week.
Analysing communication to predict clinical outcomes
Research from the team also illustrated the importance of using natural language processing (NLP) in clinician-patient interactions. They found that NLP could be extended to treatment for depression and anxiety, and discovered features of the language used in cognitive behaviour therapy (CBT) that could help automate diagnosis and predict recovery.
Working in collaboration with IESO Digital Health Ltd, the team further developed their NLP methods and applied them to their online typed service for cognitive behavioural therapy for depression and anxiety (available via the NHS).
This approach has proven effective for the automatic diagnosis of the severity of the condition, the prediction of the recovery at the end of the treatment, and to determine the quality of the therapist. To date, approximately a million patients have been treated in both the UK and US.