After a stay in the intensive care unit, around one in four patients develops symptoms of post-traumatic stress disorder (PTSD). At the same time, there is often a gap in care between discharge from hospital and further care in the family doctor's practice. Our study investigates whether an AI-supported language assistant can support this gap. The basis is Narrative Exposure Therapy (NET), a scientifically well-researched method for the treatment of trauma sequelae, which is used here in a particularly brief form. The theoretical background is the "Generalized Unsafety Theory of Stress", according to which persistent feelings of insecurity contribute significantly to the development and maintenance of chronic stress and PTSD symptoms.

In a first step, we are working with GPs, psychologists and patients to test whether the voice assistant is technically reliable and easy to use. This will be followed by a multicentre, randomized pilot study with around 75 participants. The intervention comprises eight short conversations of around seven minutes each in the GP practice over a period of two to three weeks. We are investigating whether this reduces stressful, not yet fully developed PTSD symptoms. In addition, we record perceived uncertainty, stress, anxiety and depression symptoms.

We expect that the AI-supported brief intervention will contribute to a noticeable reduction in PTSD symptoms and at the same time reduce feelings of persistent insecurity and chronic stress. The study is the first to combine an AI-based conversation offer with a modern stress theory model. However, due to the limited number of participants and possible reservations about AI in mental health care, the results should be seen as initial, indicative findings.