Using an innovative approach, researchers at the LOEWE research centre DYNAMIC want to develop a fundamentally improved understanding of mental illnesses and their treatment. To this end, the research project, which was launched in 2024, organized a large-scale, two-day retreat in January 2025 to bring together researchers from the fields of psychotherapy, psychiatry and computer science in particular. The declared aim of DYNAMIC is to bridge the professional gap between the disciplines and make joint progress in the network analysis of mental illnesses.

The main topics of the presentations and discussions at the retreat were the potential uses of artificial intelligence (AI), the potential of machine learning in predicting the course of mental illness and the innovative research of the participating scientists at TU Darmstadt, whose focus is on research into machine learning and, in particular, natural language processing (NLP), i.e. the AI-supported analysis and evaluation of human language.

A central role at the conference was played by AI researcher Dr. Hiba Arnaout from LOEWE-DYNAMIC, who emphasized the importance of AI in her presentation, as it is fundamentally changing the world – and thus also science and its “rules of the game”. Numerous Nobel Prizes in 2024 went to AI researchers from a wide range of disciplines such as physics, chemistry and others. But corresponding publications have also become significantly more frequent in the field of psychiatry and clinical psychology. Around 55% of all publications in 2023 and 2024 in these two disciplines were published with contributions from AI researchers. According to Dr. Arnaout, the quality of contributions in the field of mental health with the participation of AI scientists has unfortunately often varied so far, as they are dependent on close interdisciplinary cooperation in order to produce practice-relevant results, which is not always given or possible.

Dr. Arnaout also described specific areas of application for the use of AI methods in clinical psychology and psychiatry. They can already be used today, for example, to summarize the most important findings of a therapy session and significantly reduce the time required by patients and therapists. In addition, specialized models are used for emotion recognition during therapy or a hospital stay and thus help to assess the further course of a therapy session and an illness.

Such information is invaluable for therapists. Finally, AI-analyzed speech data can also help to predict relapses in certain mental illnesses. Dr. Arnaout cited schizophrenia as an example, where the first studies on predictability based on patients‘ statements on social media have already been carried out – provided, of course, that they have given their prior informed consent.

In her presentation, Dr. Arnaout outlined the enormous potential of AI-supported speech and text analysis for all areas of clinical psychology and psychiatry: from relapse prediction and progression diagnostics to increased efficiency in the documentation of therapy sessions, and concluded her presentation with a concrete example of interdisciplinary collaboration between AI researchers and experts from the field of mental health, which could also serve as a model for the LOEWE research centre DYNAMIC. The follow-up discussion already revealed points of contact and great mutual interest in concrete cooperation in the research projects.

As a result, the retreat and the exchanges that took place there have once again immensely increased the hope for innovative research results and the anticipation of joint research work in the coming years.

Alexander Henß and Max Berg