Human–AI interaction · quantitative methods · work and organizational psychology
Research
Research lines
1 · AI in clinical decision-making
How do clinicians, trainees, and patients evaluate, trust, and adopt AI-enabled clinical decision support – and what does it take for these tools to reach practice?
2 · AI in research and learning
When do researchers and students use and trust generative AI? Does AI-based measurement meet psychometric standards, and is trust in AI calibrated to its actual performance?
AI in clinical decision-making
AI-enabled clinical decision support systems (AI-CDSS) promise more precise diagnoses and treatment decisions, but few of them are used in routine care. Together with colleagues at LMU Munich, TH Augsburg, and University College London, I study this implementation gap from several angles and with multiple methods:
- What is being built and what reaches the market? Systematic reviews of AI-enabled precision psychiatry patents and of regulated AI decision support products for mental healthcare, including their regulatory status and evidence base.
- Why do clinicians use AI – or not? A meta-analysis of the predictors of clinicians’ intention to use AI-CDSS based on the Unified Theory of Acceptance and Use of Technology, a cross-sectional study with prospective psychotherapists, an international mixed-methods study with mental health practitioners, interviews with ophthalmologists, and a study of facilitators and barriers in psychotherapy.
- How does AI change decisions and trust? Experiments on patients’ trust when clinicians use AI, on preferences for AI-based versus dermatologist skin cancer screening, on AI advice in face matching, and on AI-enabled virtual patients in psychotherapy training.
For a non-technical overview, see In-Mind article Doctor, meet AI.
AI in research and learning
Generative AI has entered the research process itself – from literature searches and coding to data analysis and writing. This raises two kinds of questions.
Who uses AI, and why? I study the predictors of students’ AI chatbot use in an intensive longitudinal design that separates within-person from between-person effects, and when researchers trust generative AI.
Can we rely on it? My current work asks whether AI-based measurement – in particular large language model coding of open-ended and behavioral data – meets the psychometric standards the field requires, and where machine learning genuinely improves prediction in dense intraindividual data. A related focus is trust calibration: what matters is not how much people trust AI, but whether they follow correct output and reject incorrect output. I am also interested in how methods training needs to change when students can generate code and analyses with AI (see Teaching).
Evidence synthesis and quantitative methods
I develop and apply advanced quantitative methods across all of my projects:
Meta-analysis and evidence synthesis
Lead author of three published meta-analyses – on thriving at work (Journal of Organizational Behavior), students’ career exploration (Journal of Vocational Behavior), and clinicians’ intention to use AI (JMIR) – co-author of a meta-analysis on human energy at work, and first author of a meta-analysis on future work self salience (under review). Random-effects and multilevel models, meta-regression, relative weights, and publication bias diagnostics.
Latent variable and multilevel models
Structural equation modeling (Mplus, lavaan), latent profile analysis, latent change score models, multilevel and polynomial regression, within- versus between-person decomposition, and measurement invariance.
Design and power
Co-author of a tutorial on simulation-based sample-size planning for generalized linear mixed models (Advances in Methods and Practices in Psychological Science); randomized online experiments, conjoint and vignette designs, and advice-taking paradigms.
Machine learning and computational methods
Python, R, and SQL; supervised and unsupervised learning (Le Wagon Data Science traineeship); Bayesian modeling; natural language processing; and LLM-based coding of text with systematic evaluation of output quality.
Thriving, careers, and well-being at work
My PhD at the University of Groningen focused on adaptation and thriving during key vocational transitions. This line of work continues today:
- Thriving and energy at work – a meta-analysis on thriving at work, recognized as the most-cited article in the Journal of Organizational Behavior (2020); a four-wave study on vitality, learning, and health; and a meta-analysis on human energy.
- Careers and future work selves – students’ career exploration, career planning and career-related worry, a meta-analysis on future work self salience, and the development of an AI-specific future work self scale with Karoline Strauss (ESSEC Business School).
- Entrepreneurship – how entrepreneurs appraise and cope with errors and how financial stress relates to their intention to quit.
Open science
I preregister studies, share materials, data, and code (see GitHub and the links on the publications page), and support Registered Reports as a recommender at PCI RR. As a BITSS Catalyst, I develop open science training material, and I am a member of the LMU Open Science Center.
Interested in collaborating? Get in touch.
Last updated: October 2026