Horizon CDT Research Highlights

Research Highlights

User Trajectories on Digital Mental Health Platforms

  Gregor Milligan (2021 cohort)

This PhD aims to explore how natural language processing techniques can be applied to data generated by users on digital counselling and support platforms to understand and predict the trajectory of the platform users. A trajectory can be defined as the steps taken by a participant in a service or experience to get to a specific outcome (Benford et al., 2009). Within the context of a counselling platform, these would be the steps taken to achieve a valuable outcome for the user.

Sefi and Hanley (2012) have previously explored the role of standardised measures in understanding the needs and goals of service users on digital support platforms (Kooth). Kooth is a web-based counselling and support platform for children and young people (CYP) aged between 11-25 years old (Hanley et al., 2021). The needs of each user on the Kooth platform is unique, and therefore each trajectory will be different. To ensure the service user gains the maximum value from the service, it is vital that the user is presented with a personalised (idiographic) experience based on their needs and goals while using the platform. However, CYP using the Kooth service were shown to have a broader range of needs than face-to-face services, suggesting that a more personalised approach to defining the needs and goals of a service user would be appropriate.

Natural language processing (NLP) is a collection of computational techniques for learning, understanding, and producing human language content (Hirschberg & Manning, 2015). NLP techniques, including sentiment analysis, topic modelling and text summarisation, can be used to detect suicide ideation from counselling transcripts and depressive symptoms from social media data (Coppersmith et al., 2018). This research aims to apply NLP techniques to user-generated data to define personalised trajectories for service users accessing digital support platforms. The output from this research will enable a more accurate understanding of the needs and goals of the service users and allow for a more tailored experience based on the expected trajectory of the user.

Research Questions and Objectives:

  1. Determine to what extent there are different types of users on digital mental health platforms and how each user type interacts with the platform.

  1. Investigate and apply existing machine learning and statistical modelling techniques to data generated on digital mental health platforms to predict the trajectory of users (In terms of both the use of the platform and user wellbeing).

  1. Explore how natural language processing (NLP) techniques can be applied to text data generated on the digital mental health platforms to understand service user needs, trajectory, and engagement.

  1. Explore how mental health and digital wellbeing theory can validate outputs from data-driven techniques for understanding users of the platform.

  1. How can idiographic (personalised) mental health and wellbeing measures be used to support the understanding of user trajectories on digital mental health platforms?

Bibliography

Benford, S., Gabriella, G., Boriana, K., & Rodden, T. (2009). From interaction to trajectories: Designing coherent journeys through user experiences. Conference on Human Factors in Computing Systems - Proceedings, (pp. 709-718 https://doi.org/10.1145/1518701.1518812).

Coppersmith, G., Leary, R., Crutchley, P., & Fine, A. (2018). Natural Language Processing of Social Media as Screening for Suicide Risk. Biomedical Informatics Insights, 10, 1-11 https://doi.org/10.1177/1178222618792860.

Hanley, T., Sefi, A., Grauberg, J., Prescott, J., & Etchebarne, A. (2021). A theory of change for web-based therapy and support services for children and young people: Collaborative qualitative exploration. JMIR Pediatrics and Parenting, 4(1), https://doi.org/10.2196/23193.

Hirschberg , J., & Manning, C. D. (2015). Advances in natural language processing. Science, 349(6245), 261-266 https://doi.org/10.1126/science.aaa8685.

Sefi, A., & Hanley, T. (2012). Examining the complexities of measuring effectiveness of online counselling for young people using routine evaluation data. Pastoral Care in Education, 30(4), 49-64 https://doi.org/10.1080/02643944.2011.651224.