Project MentalWords: Efficient collection of text data in a clinical environment and linguistic analyses

The MentalWordsproject is investigating what happens to linguistic expression when someone is affected by mental health problems. But first of all, it needs data.

Important questions, less data

Natural language processing (NLP) is increasingly being used to better understand mental health. This involves automatically analysing texts such as forum posts or conversation transcripts in order to identify indications of moods, stress or clinically relevant symptoms. Such methods can support doctors and therapists by providing additional information or revealing patterns that are difficult to recognise with the naked eye. The long-term aim is to help patients earlier and in a more targeted manner.

However, a major problem in this area of research is the lack of suitable data. Clinical data in particular, i.e. speech or text samples from real patients undergoing treatment, are rarely accessible as they are strictly confidential and sensitive. Researchers therefore often have to rely on public sources such as social media, which are not always reliable or representative. Few studies have direct access to real clinical data, which makes the development and testing of models considerably more difficult. A review study showed that social media data was used in over 80% of studies [1].

Promising initial approaches

The team at BFH has already addressed the issue in previous work: for example, how text analysis can be used to recognise the first signs of burnout. To this end, the research team analysed anonymous posts by people talking about their experiences online [2] and collected anonymous texts using a survey [3]. The aim was to find patterns in the language that could indicate particular stress or exhaustion.

The study has already shown initial positive results: The developed method [2] recognises burnout clues in texts quite reliably. However, it should be noted that these results still need to be further verified, especially with clinical data and in practical fields of application.

The MentalWords project

The MentalWords project is now taking up this thread. In close co-operation between the BFH School of Engineering and Computer Science and the BFH School of Health Professions, it is being carried out together with the University of Bern (University Psychiatric Services Bern) and the clinical partner Privatklinik Meiringen. The aim of the project is to develop innovative approaches in the field of computational linguistics in psychiatry and to test them in practice. The project will run for four years and is funded by the Swiss National Science Foundation.

The focus is on the development of a data collection protocol that can be integrated as smoothly as possible into the clinical workflow in order to minimise the effort for the clinical partner in interdisciplinary research collaborations in the long term. The transcribed data will then be analysed to explore differences in written expression between patients with burnout, depression, anxiety and a healthy control group. The aim is to combine scientific findings from different disciplines with clinical expertise in order to create concrete insights for the psychiatry of the future.

An interdisciplinary and translational team

An interdisciplinary and translational team is very relevant for such a project. The project team is made up of experts from the fields of health sciences, medicine and computer science/computational linguistics and is working on the implementation in an interdisciplinary manner. The project is led by Prof. Dr Mascha Kurpicz-Briki from the Applied Machine Intelligence research group at the School of Engineering and Computer Science, together with Prof. Dr Thomas J. Müller from the University of Bern (University Psychiatric Services Bern) and the Privatklinik Meiringen, in close collaboration with Prof. Dr Christoph Golz from the BFH School of Health Professions. The two research groups have already worked together successfully on various other projects.

Science communication involved

The Institute of Design Research at Bern University of the Arts (HKB) is also supporting the project in communicating the research. The project objectives are to be explained in a way that is understandable to a wide audience in order to achieve a broad impact.

Project website:
https://www.bfh.ch/de/forschung/forschungsprojekte/2025-790-622-342/

References

[1] Zhang, T., Schoene, A. M., Ji, S., & Ananiadou, S. (2022). Natural language processing applied to mental illness detection: a narrative review. NPJ digital medicine, 5(1), 46. https://www.nature.com/articles/s41746-022-00589-7.pdf

[2] Merhbene, G., Nath, S., Puttick, A. R., & Kurpicz-Briki, M. (2022). BurnoutEnsemble: augmented intelligence to detect indications for burnout in clinical psychology. Frontiers in big Data, 5, 863100.

[3]Kurpicz-Briki, M., Merhbene, G., Puttick, A., Souissi, S. B., Bieri, J., Müller, T. J., & Golz, C. (2024). Using Natural Language Processing to find Indication for Burnout with Text Classification: From Online Data to Real-World Data. arXiv preprint arXiv:2409.14357.

Dr Mascha Kurpicz-Briki is Professor of Data Engineering at the Institute for Data Applications and Security IDAS at Bern University of Applied Sciences, and Deputy Head of the Applied Machine Intelligence research group. Her research focuses, among other things, on the topic of fairness and the digitalisation of social and community challenges.

Christoph Golz is Head of Healthcare Innovation Field - Human Resources Development at the Bern University of Applied Sciences Health. His research focuses on the future-oriented and demand-driven development of health care.

Prof. Jimmy Schmid is a communication designer and a member of the management team of the Institute of Design Research IDR at Bern University of the Arts HKB. He coordinates the two research fields of Environmental Communication Design and Knowledge Visualisation.
He is head of the part-time postgraduate programme MAS Signage.
He is also a guest lecturer and expert at various national and international universities, colleges and institutions and the author of various signage technology articles and signage technology study assignments as well as a consultant on signage technology issues (juries, competitions, agency evaluations).

Shakir Sultanov is a fullstack developer who works as an intern in the Institute for Data Applications and Security at the Bern University of Applied Sciences.

Leonie Roos is a qualified nurse MScN and works as a doctoral candidate in the field of Health Care Innovation and Human Resource Development at the Department of Health of the Bern University of Applied Sciences.

Prof. Dr. Thomas J. Müller is the Medical Director of the Private Clinic Meiringen and a specialist in psychiatry and psychotherapy, focusing on ADHD, autism, and the influence of environmental factors on mental health. He is also actively involved in research and teaching – particularly at the University of Bern – and maintains an international network within the scientific community.

Nicolo Bernasconi is a visual designer, communication designer, and research associate at the Institute of Design Research (IDR) of the Bern University of the Arts (HKB). His research focuses on knowledge visualization, environmental communication design, and signage systems. In addition, he is active in professional committees — including participation in juries, curatorial boards, and project management — and combines strategic, conceptual, and experimental approaches in his work to further develop design processes and visual communication.