Research
AI ethics and meta-research
Our research focuses on AI ethics, meta-research and responsible research practices in health AI.
Research areas
Select an area to highlight its related projects below.
Meta-research and research quality
We study how AI research is conducted and evaluated, with an emphasis on robustness, transparency and the evidence supporting claims about performance. Methods include systematic reviews and methodological analyses.
For example, our review of AI-based scoring systems in intensive care examined the frequency of external validation and changes in model performance on external data.
As of October 2026, two systematic reviews on robustness practices in medical imaging and mental health have been finalised for submission.
AI development, deployment, and evaluation
We study methods for developing, deploying and evaluating healthcare AI, with attention to robustness, transparency and bias.
This includes model development and validation, assessment of data-related bias, and ethical evaluation in the intended context of use.
Implementation of AI ethics
We investigate how ethical principles are interpreted and implemented during AI development. AI ethics is principle-based. But for specific projects, these principles need to be contextualized and operationalized. Methods include qualitative interviews, co-creation workshops and ethical co-design.
Recent studies examine the institutional and practical constraints on ethics implementation and the formulation of development requirements with stakeholders.
Responsible research practices
We develop research and services to support responsible AI development at Charité. We also contribute to projects on AI literacy and the use of AI in education.
Current and upcoming grants and projects
Related to selected area
MOTHER
Maternal Health, Obesity Trajectories and Health Equity Research
BMFTR · Funded
Planned start: January 2027
Planned duration: 24 months
Using IDEFICS/I.Family cohort data, MOTHER will examine the added value of maternal and early-life information for modelling obesity and metabolic trajectories in girls. Our group will develop and validate AI models and methods to identify and reduce data-related bias.
Related publications
No publications to date.
Related to selected area
ANONY-MED 2
June 2026–May 2028 · Funded
BMFTR · Digital. Sicher. Souverän.
ANONY-MED 2 continues ANONY-MED. We extend the existing framework for evaluating the trustworthiness of synthetic medical data generation by adding quantitative evaluation criteria.
Project description (external link) ↗
Related publications
No publications listed yet.
Related to selected area
TEF-Health
January 2023–December 2027
European Commission, Digital Europe Programme, and national co-funding
Grant 101100700
Our group contributes assessment procedures addressing transparency, robustness and other ethical requirements for medical AI. The wider consortium develops testing and experimentation facilities for health AI and robotics.
Related to selected area
STRATIF-AI
May 2023–April 2027
European Union, Horizon Europe · Grant 101080875
Our group is responsible for the definition of ethical and trustworthy AI development and assessment requirements within this project. The consortium studies AI and digital twins across stroke prevention, treatment and rehabilitation.
Project description (external link) ↗
Related publications
- No publications listed yet
Ongoing programmes and services
Related to selected area
Responsible Translation
Ongoing research programme · BIH QUEST
This ongoing meta-research programme examines trustworthiness, robustness, usefulness and transparency in data- and algorithm-driven biomedical research. Its approach combines conceptual and methodological work, narrative and systematic reviews, and stakeholder interviews to inform recommendations for research practice.
Related to selected area
Trustworthy AI in Healthcare Lab
I lead the Z-Inspection® Trustworthy AI in Healthcare Lab at QUEST. The lab brings together experts from healthcare, computer science, ethics, law and other relevant disciplines to assess ethical issues in specific AI applications. Assessments use the published Z-Inspection® approach and consider the application context and stakeholders involved.
Related publications
Completed grants and projects
Related to selected area
VALIDATE
May 2022–April 2026 · Completed
European Union, Horizon Europe · Grant 101057263
Our group was responsible for the definition of ethical and trustworthy AI development and assessment requirements within this project. The consortium developed an AI system for stratification of acute stroke patients.
Project description (external link) ↗
Related publications
Related to selected area
ANONY-MED
December 2022–December 2025 · Completed
German Federal Ministry of Education and Research (BMBF) and European Union
Our group was responsible for a trustworthy-AI framework covering ethics, robustness and transparency. The project as a whole investigated privacy-preserving approaches to medical data use, including synthetic data and encrypted analysis.
Project description (external link) ↗
Related publications
No publications to date.
Related to selected area
INTER-AGENT
December 2021–November 2025 · Completed
BMBF and State of Berlin · Federal–state initiative “Künstliche Intelligenz in der Hochschulbildung”
The project addressed AI-supported tutoring and interdisciplinary AI education for students in health professions. Our contribution concerned development of an ethical framework for trustworthy educational AI.
Project description (external link) ↗
Related publications
No publications to date.