Modelling patient mental, behavioural and somatic experiences using Natural Language Processing to improve risk detection of suicidal behaviour and self-harm

Lead Supervisor
Dr Rina Dutta
Senior Clinical Lecturer in Psychological Medicine
Department of Psychological Medicine, Institute of Psychiatry, Psychology & Neuroscience (IoPPN), King’s College London

Dr Sumithra Velupillai
Lecturer in Applied Health Informatics, King’s College London

Industrial Partner

Project Details

Within mental healthcare, patient safety is difficult to define because of the often interrelated understanding of disorder and behaviours.  For example, suicide and self-harm are unsafe behaviours not diagnoses, making their accurate detection and modelling from electronic health records a challenge.

Clinicians are taught that when recording the presenting complaint in the psychiatric history they should give a brief description in the patient’s own words.  Although some patients use symptom terms, (e.g. I feel suicidal), many describe their experiences in their own words (e.g. I feel like ending it all). For accurate diagnostic formulation and appropriate treatment planning, clinicians condense and interpret the information they glean from the patient interaction, often focusing on symptoms and risk issues.  Clinical academics generally agree more research into suicide prevention is needed, but service users are more concerned with the challenges of experiencing suicidal ideation, intentions and self-harming behaviour, which impact on daily living, well-being, relationships and social issues.

This PhD project has a novel goal in that it will use the latest techniques in natural language processing (NLP), machine learning and text analytics to model patient experiences. Modelling patient experiences involves developing novel NLP approaches that are able to capture complex expressions and idioms (e.g. for suicidality: don’t want to wake up in the morning; for self-harm: just want to release the tension), for which recent advances in language representation models would be very relevant.

Furthermore, the project aims to accelerate the translation of these findings into research to optimise outcomes that are important to patients. For example, this work could be used for improved patient recruitment in clinical trials, by better identification based on patient experiences. There are also clear opportunities for transdiagnostic studies, as the focus is on experience rather than symptoms or diagnostic categories.

DRIVE-Health is associated with one of the UK’s leading centres in clinical NLP, the NIHR Maudsley Biomedical Research Centre (BRC). The BRC routinely runs more than 80 different NLP applications over 35 million documents, in order to support health research in mental health. The BRC has a large and active NLP, data science, and health informatics research groups.

The project is sponsored by AIMES, a cloud service provider and data analytics company with contracts throughout the health sector. The student will benefit from opportunities to explain their work to a commercial audience, which will bring insights in how their work may be generalised, and will benefit from potential opportunities to validate methods with AIMES health clients.  


Data from the Clinical Records Interactive Search (CRIS) database, which contains anonymized EHRs from the South London and Maudsley (SLaM) NHS Foundation Trust and has ethical approval for research use (Oxford REC C, reference 08/H0606/71+5) under an extensive governance model will be central to the initial development of NLP algorithms. Opportunities for further development and evaluation on other similar data in other NHS trusts, e.g. MerseyCare will be actively explored.


Mental health, natural language processing, NLP, EHR, electronic health records, suicidal behaviour, self-harm