JGJürgen Gallinat

Jürgen Gallinat

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Jürgen Gallinat has a total of 319 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Hybrid Clinician-in-the-Loop Monitoring System for LLM Conversational Agents

    by Chaewon Yun, Rodrigo Bermudez Schettino, Brooke C. Schneider, Simone Kühn, Jürgen Gallinat, Ronald Ashri, Iyad Rahwan

    Large language models (LLMs) are increasingly deployed as conversational agents, owing to their ability to sustain coherent interactions. However, the stochastic nature of LLMs and well-documented risks such as hallucinations and unsafe outputs pose significant challenges for ensuring safe deployment. We introduce a novel monitoring system that we designed and implemented to tackle these challenges in deployments of LLM conversational agents. Our system combines expertise from psychiatry and computer science to support risk detection and multi-stakeholder decision-making. The system consists of two main modules: continuous regression testing and real-time monitoring. Regression testing leverages multi-turn synthetic conversations to evaluate the safety of LLMs in deployment and runs automatically in a continuous integration pipeline. Real-time monitoring incorporates both manual and automated evaluation of user-model interactions, including efficient evaluation from mental health professionals via a web-based interface. To demonstrate how our system works in practice, we conducted a preregistered online behavioral experiment ‘AI Safety Challenge’ as a case study, which demonstrates the feasibility of our system in live operation and suggests a modest efficiency gain under a fixed expert review budget. Our monitoring system functions as a platform where different types of expertise can be combined. Clinical expertise informs dataset design, evaluation procedures, and real-time monitoring, complementing computational safety metrics to deliver a robust monitoring infrastructure. Best practices and state-of-the-art methods from computer science are employed to build a robust and scalable architecture. Executable code 1 and datasets 2 are provided to support reuse, extension, and reproducibility in future applications. Warning: This paper includes model-generated outputs that may be offensive or distressing, including references to self-harm.

  • Open Access

    Virtual Exposure With Response Prevention for Obsessive-Compulsive Disorder: Randomized Controlled Trial

    by Lara Rolvien, Lena Jelinek, Luzie Lohse, Steffen Moritz, Lisa Borgmann, Jürgen Gallinat, Simone Kühn, Franziska Miegel

    Background: Despite being highly effective and guideline-recommended for obsessive-compulsive disorder, exposure with response prevention is often underused due to barriers from both therapists and patients. Virtual reality may help address these barriers. Objective: This study aimed to assess the efficacy and acceptance of virtual exposure with response prevention (VERP) in patients with contamination obsessive-compulsive disorder (conOCD) and checking obsessive-compulsive disorder (checkOCD) and to identify factors influencing treatment success. Methods: Within a randomized controlled trial, 80 participants (40 with conOCD and 40 with checkOCD) were allocated to an intervention group, which received outpatient care as usual (CAU) supplemented by 6 weekly VERP sessions (VERP group), or to a control group receiving CAU alone (CAU group). Assessments occurred at baseline (t0), postintervention (t1), and at 3-month follow-up (t2). The reduction in obsessive-compulsive symptoms, as measured with the Yale-Brown Obsessive-Compulsive Scale, served as the primary outcome. Results: Overall, obsessive-compulsive symptoms showed no greater improvement in the VERP group compared to the CAU group from t0 to t1 or t2 (η p 2 =0.003-0.014). In exploratory subsample analyses, we observed a trend toward an interaction effect, showing greater reduction in obsessions among participants with checkOCD from t0 to t1 ( F 1,52 =3.91; P =.05; η p 2 =0.070) and greater anxiety reduction among participants with conOCD from t0 to t2 ( F 1,45 =4.11; P =.05; η p 2 =0.084). Exploratory moderation analysis suggested that adding VERP to CAU was more effective for those in the VERP group who had not prematurely discontinued prior psychological treatment and those with checking behavior. Sense of presence was only moderate, unrelated to outcome, and patient satisfaction with VERP was positive overall. Conclusions: While overall symptom reduction did not significantly differ between groups, exploratory analyses suggested trend-level differences in treatment response across obsessive-compulsive disorder subtypes and patient characteristics. These findings should be interpreted cautiously and are best viewed as hypothesis-generating.

  • Open Access

    Coping, depressive and anxiety symptoms over three and a half years of the COVID-19 pandemic: a cross-lagged panel network analysis

    by Laura Kenntemich, Leonie von Hülsen, Laura Heinrich, Jonas Petter, Levente Kriston, Jürgen Gallinat, Ingo Schäfer, Annett Lotzin

  • Open Access

    Informing the redesign of psychiatric seclusion rooms: a mixed-methods pre-evaluation with individuals with lived experience

    by Leonie Ascone, Candelaria Mahlke, Nour Tawil, Larissa Samaan, Martin Frisch, Lena Nugent, Rebecca Nixdorf, Florian Börncke, Daniel Lüdecke, Rabea Fischer, et al.

    Background: In acute psychiatric inpatient settings, where perception is altered and emotional vulnerability is heightened, many facilities use coercive seclusion rooms for safety. This practice has been increasingly criticized for its psychological impact. While architectural guidelines emphasize human-centered and biophilic design to support well-being, empirical evidence on how design features in this specific setting affect patients remains limited. Methods: This mixed-methods study examined how individuals with lived experience of coercive isolation in seclusion rooms ( N = 30) perceived different wall designs. The participants viewed ten digitally rendered images of a to-be-redesigned seclusion room, including furniture, varying in wall design (six nature-themed images, three wall colors: blue, green, beige; and a white empty control room) and rated each on restfulness, stress, liking, and overwhelm. This was followed by qualitative interviews on needs and design preferences. Results: Nature-themed wallpapers, especially an image of grass-covered dunes by the sea, and blue and green wall colors were rated as more restorative and less stressful than both the empty white control room and a beige-painted comparison room, or a more complex, ‘rough’ wilderness nature image. Qualitative findings emphasized calm, homelike, nature-themed, and controllable environments, orientation, more transparent and empathetic communication, and respectful care. Conclusions: Individuals with lived experience respond more favorably to blue and green color schemes and non-complex, calm nature imagery, compared to more neutral or sterile low-stimulation designs, challenging long-standing assumptions that sensory deprivation best supports de-escalation in seclusion. The study further illustrates the feasibility and ethical value of participatory pre-evaluative research to inform design. Trial registration: The study and main analyses were pre-registered 2024/08/22 at aspredicted.org (#187275) https://aspredicted.org/9ht9-ckwn.pdf . Clinical trial number: Not applicable.


  • Open Access

    Developing brains and changing worlds: Macroenvironmental changes and their association with brain development

    by Juan F. Quinones, Tobias Banaschewski, Arun L.W. Bokde, Sylvane Desrivières, Herta Flor, Hugh Garavan, Penny Gowland, Antoine Grigis, Andreas Heinz, Herve Lemaitre, et al.


  • Open Access

    Exploring the concept of a “friendly relationship to death” in geriatric patients

    by Zoi Raboti, Alexander Rösler, Reinhard Lindner, Jürgen Gallinat, Brooke C. Schneider

    Background: Positive attitudes toward death are common among geriatric patients, yet their distinction from mental illness and suicidality remains insufficiently understood. This study sought to characterize the concept of a “friendly relationship to death” (FRD), including its frequency in a geriatric sample and identifying variables from the geriatric assessment associated with a FRD. Methods: A cross-sectional study was conducted in a geriatric unit, surveying N = 99 patients aged 65 years and older. Attitudes toward death were assessed using The Multidimensional Orientation Toward Dying and Death Inventory ( MODDI-F ) , supplemented by one question addressing a FRD (“I have a friendly relationship to death.”), as well as eight additional questions on attitudes toward death, dying and religiosity/spirituality . Furthermore, depression, suicidality, disease burden, frailty, and loneliness were evaluated using validated questionnaires. Results: A total of 70% of patients endorsed a FRD. A FRD was associated with greater acceptance of one’s own dying and death, lower fear and rejection of death, greater religiosity/spirituality, and a stronger sense of “satisfied hunger for life.” A FRD was not associated with depression, suicidality, overall disease burden, frailty or loneliness. Conclusion: Among geriatric patients, a FRD appears to prevail, extending beyond a neutral acceptance of death’s inevitability and existing independently of suicidality, depression, frailty, or disease burden. Understanding a FRD may help contextualize older individuals’ acceptance of aging and could impact geriatric treatment outcomes.

  • Transformation of MetaHumans for a Generic XR Workflow

    by Christian Zimmer, Lucie Kruse, Lara Rolvien, Jürgen Gallinat, Christian Geiger


  • Open Access

    IVAs' Future in Therapy: Investigating the Impact of Intelligent Virtual Agents in VR on Psychological Self-Help Exercises and User Satisfaction

    by Sebastian Rings, Lucie Kruse, Catharina Rudschies, Lara Rolvien, Gesche Schauenburg, Jürgen Gallinat, Ingrid Schneider, Frank Steinicke


  • Open Access

    Utilisation of alcohol-related treatment after a first alcohol use disorder diagnosis in Hamburg, Germany

    by Jakob Manthey, Kilian Huß, Sven Buth, Ludwig Kraus, Anna Schranz, Carolin Kilian, Jürgen Gallinat, Ingo Schäfer, Bernd Schulte

    Background: A variety of treatment options for people with alcohol use disorder (AUD) exist. Surveys estimate that 1 in 10 people with AUD utilise treatment, but real-world treatment pathways remain covert. This data-linkage study seeks to characterise treatment utilisation patterns to identify gaps in treatment access and delivery in Germany. Methods: Linking individual-level data from three sources (statutory health insurance, pension funds, outpatient addiction care services) identified seven alcohol-related treatment types delivered in outpatient (brief psychiatric consultation; formal psychotherapy; pharmacotherapy; low-threshold counselling), inpatient (standard, somatic inpatient treatment; intensive inpatient treatment with somatic and psychosocial care), or either of the two settings (long-term rehabilitation treatment) during 2016–2021. For patients with a new AUD diagnosis (ICD-10: F10.1–9), treatment utilisation over 24 months was recorded and patterns were identified using latent class analyses. Results: Of n = 9,491 patients with a new AUD diagnosis, 30% utilised at least one alcohol-related treatment type. Treatment utilisation was associated with younger age, female sex, unemployment, German nationality, and lower physical comorbidity. Among treatment entrants, nearly half received only brief psychiatric consultation. A similar share of patients utilised standard or intensive inpatient treatment; the latter occasionally followed by rehabilitation treatment. Formal psychotherapy, low-threshold counselling, and pharmacotherapy were rarely utilised and were mostly used in conjunction with other treatments. Conclusions: The real-world utilisation of alcohol-related treatments contrasts with existing guidelines, as most patients with diagnosed AUD do not receive adequate care. Structural and social barriers should be minimised to ensure healthcare provision for those affected.


  • Open Access

    Development and internal validation of a gradient-boosted trees model for prediction of delirium after surgery and anesthesia (the BioCog study)

    by Florian Lammers-Lietz, Levent Akyuez, Diana Boraschi, Friedrich Borchers, Jeroen de Bresser, Sreyoshi Chatterjee, Marta M. Correia, Nikola M. de Lange, Thomas Bernd Dschietzig, Soumyabrata Ghosh, et al.

    IMPORTANCE: Postoperative delirium (POD) is a multietiological condition and affects 20% of older surgical patients. It is associated with poor clinical outcome and increased mortality. OBJECTIVE: We aimed to develop and validate a risk prediction algorithm for POD based on a multimodal biomarker database exploiting preoperative data (predisposing factors) and procedural factors as well as perioperative molecular changes associated with POD (precipitating factors). DESIGN: BioCog is a prospective cohort study conducted from November 2014 to April 2017. Patients were followed up for seven postoperative days after surgery for POD. Gradient-boosted trees (GBT) with nested cross-validation was used for POD prediction. SETTING: Patients aged ≥65 years were enrolled at the anesthesiologic departments of two tertiary care centers. EXPOSURE: All patients underwent surgery with an expected duration of at least 60min. Clinical, neuropsychological, neuroimaging data and blood were collected and clinically well established as well as non-established biomarkers (e.g., gene expression profiling) were measured pre- and postoperatively. MAIN OUTCOME: POD according to DSM 5 until the seventh postoperative day RESULTS: 184 of 929 (20%) patients experienced POD. A GBT algorithm using both preoperative data, characteristics of the intervention and postoperative changes in laboratory parameters achieved the highest area under the curve (0.83, [0.79; 0.86]) with a Brier score of 0.12 (0.12; 0.13). CONCLUSIONS AND RELEVANCE: Models combining predisposing factors with precipitating factors predict POD best. Non-routine laboratory data provide useful information for POD risk prediction, providing relevant results for future studies on the molecular factors of POD. In addition, possibly relevant molecular mechanisms contributing to the development of POD were identified, mostly indicating a dysregulated postoperative immune response. This study constitutes the basis for future hypothesis-driven analyses or implementation of prediction expert system for clinical practice.