Project ID NS-MH2026_37

ThemeNS-MH

Co Supervisor 1A Dr Sarah Morgan Faculty of Life Sciences & Medicine, School of Biomedical Engineering & Imaging Sciences, Department of Biomedical ComputingEmail

Co Supervisor 1B Dr Hector Menendez Benito Faculty of Natural, Mathematical & Engineering Sciences, Department of InformaticsEmail

Third Supervisor Dr Kelly Diederen

Can digital communication messages be used to predict later psychotic illness?

There is pressing clinical demand for tools that can predict the course of psychotic disorders at the individual level, enabling earlier, preventive interventions. Our prior work showed that Natural Language Processing (NLP) markers of altered language use have significant predictive power. However, most studies focus on group-level differences in NLP metrics, rather than modelling longitudinal changes within individual patients.

This project aims to bridge that gap by using NLP markers extracted from patients’ digital messages to predict later psychotic illness. Our key objectives are:

Year 1: Develop an app to extract anonymised NLP features from previously sent text or Whatsapp messages, which will be tested using participants from the general population.

Year 2: Use the app to collect data from patients with psychotic disorders, and link to clinical electronic health record data (year 2).

Year 3: Assess whether the NLP features extracted can predict future clinical events (e.g. hospitalisations).

People with lived experience of psychosis will be involved throughout and the student will receive training in co-creation of Digital Health research. The supervisors have substantial experience of patient and public involvement, e.g. co-designing an app to record speech data with psychosis patients.

If successful, the tools developed could be adapted for other conditions, e.g. depression or Alzheimer’s.

This project would suit a student with good programming skills, a first degree in e.g. Computer Science, Engineering, Mathematics, Physics, Neuroscience or Medicine, and an interest in AI for mental health applications. The student will develop expertise in NLP, AI, digital health and psychosis.

Rotation project: The student will assess the computational demands of extracting various NLP measures and determine which measures provide complementary, non-redundant signals, using large, pre-existing datasets (e.g. ENTER, N=5,000). This will inform which features should be prioritised in the PhD. The student will develop expertise in NLP and psychosis.

Representative Publications

1) Morgan et al., “Natural Language Processing markers in first episode psychosis and people at clinical high-risk”, Translational Psychiatry, 11, 1-9, 2021, https://doi.org/10.1038/s41398-021-01722-y. 2) Nettekoven et al., “Semantic speech networks linked to formal thought disorder in early psychosis”, Schizophrenia Bulletin, 49, S142, 2023, https://doi.org/10.1093/schbul/sbac056. 3) Olah et al., “Towards a scalable approach to assess speech organization across the psychosis-spectrum -online assessment in conjunction with automated transcription and extraction of speech measures”, Translational Psychiatry, 14, 156, 2024, https://doi.org/10.1038/s44277-025-00034-z.

1) Barnard, P., Bautista, J. R., Dakhama, A., Farahi, A., Laos, K., Liu, A., & Menéndez, H. D. (2024, October). Responsible MLOps Design Methodology for an Auditing System for AI-Based Clinical Decision Support Systems. In IFIP International Conference on Testing Software and Systems (pp. 217-236). Cham: Springer Nature Switzerland. https://doi.org/10.1007/978-3-031-80889-0_16 2) Okazaki, S., Plangger, K., West, D., & Menéndez, H. D. (2020). Exploring digital corporate social responsibility communications on Twitter. Journal of Business Research, 117, 675-682. https://doi.org/10.1016/j.jbusres.2019.09.006 3) Iglesias, Guillermo, Hector Menendez, and Edgar Talavera. “Improving explanations for medical X-ray diagnosis combining variational autoencoders and adversarial machine learning.” Computers in Biology and Medicine 188 (2025): 109857. https://doi.org/10.1016/j.compbiomed.2025.109857