![[Translate to English:]](/fileadmin/_processed_/9/b/csm_Firefly_Abstract_organic_flow_resembling_blood_vessels_shaped_as_a_tool__composed_of_glowing_651800_26778ea790.png "[Translate to English:]")
The AI Ethics research group at the IMDE investigates how the use of AI in medicine can be shaped in an ethically responsible manner. The focus here is on medical responsibility. A key concern of the institute is to precisely define the scope of responsibility and its limits in everyday clinical practice: what do doctors owe their patients when a decision is based on the output of a system that is not entirely transparent? What level of scrutiny can realistically be expected in day-to-day clinical practice, where does individual accountability end, and how is responsibility distributed amongst healthcare providers, institutions and manufacturers?
Closely linked to this are the normative justification for explainability requirements, the epistemic prerequisites of medical decisions and their context-dependence: clinical application areas differ in normatively relevant ways, for example with regard to time pressure and the extent of intervention.
Further areas of focus include AI-based decision support for doctors, the evaluation of AI-based biomarkers in terms of validation, generalisability and fairness, the role of patients in AI-supported processes, and the preservation of medical expertise in the face of increasingly automated procedures. The research group combines normative analysis with empirical research and contributes to professional association standards for oncology.
Projects
AI-powered Insights into Patient Decisions
Participation in clinical trials is essential for medical progress. Nevertheless, many cancer patients choose not to take part in trials or withdraw from them prematurely. The reasons for this often remain unclear, as traditional research approaches primarily reach patients who are already open to the healthcare system and research. Using artificial intelligence, the project analyses publicly available posts from internet forums, blogs and social media to systematically evaluate perspectives that have previously received little attention. The aim is to identify psychosocial, structural and communicative barriers to participation in clinical trials, as well as reasons for dropping out, and to develop new hypotheses for social science research. In the long term, the project aims to contribute to a better understanding of patient recruitment and retention in clinical trials and to develop patient-centred strategies to promote trial participation.
IMDE Contact Partners
MEDAL (MEDical Imaging AGI’s Last Exam)
The project MEDical Imaging AGI’s Last Exam (MEDAL) aims to create a next-generation benchmark to assess the true capabilities of medical imaging AIs in answering the biggest and most challenging clinical/medical imaging tasks. This would direct the research capacity of the AI community toward the questions of greatest global salience, as identified through open, worldwide public consultation. This interdisciplinary and highly international project works together with our ethics team to ensure that issues such as relevance, distributive justice, data management and transfer, and neutrality are given appropriate considerations from an ethical perspective. The project has begun in 2026 and is led by the German Cancer Research Center.
IMDE Contact Partners
ONCOverse
WP-6 – Embedded Ethics and Technology Acceptance
A Digital Care Platform for Pancreatic Cancer
The ONCOverse project (ONCOverse: Patient-Centered and Intuitive Evidence Generation for Pancreatic Cancer), funded by German Cancer Aid, is driving the development of a digital care platform to refine the treatment of pancreatic cancer and enable the best possible, individually tailored care. To this end, patients will contribute by providing their treatment data, experiences, and information about their well-being. At the same time, treating physicians will also enrich the platform. To achieve this, innovative digital technologies—including artificial intelligence—are being integrated to process the large volumes of data effectively and contribute to data-driven decision support.
Ethical and Technical Design in Work Package 6
Work Package 6 (WP-6 – Embedded Ethics and Technology Acceptance) of the project focuses on how these technologies can be designed in a responsible, secure, understandable, and user-oriented manner. In particular, the project examines the ethical challenges of human-AI interaction, AI-supported decision-making processes, the ultimate responsibility or decision-making authority of physicians, data protection, and the use of medical data.
User Perspectives and Equitable Participation
At the same time, the project empirically addresses the needs and expectations of patients, physicians, and researchers, as this combined group of stakeholders , as these consolidated stakeholder perspectives are directly incorporated into the development and design of ONCOverse technologies in line with the principles of user-centered design. Fair access to medical innovation also plays a particularly important role in this context. WP-6 examines potential barriers to access and develops recommendations on how to reduce them.
Ethics by Design for Trustworthy Technologies
WP-6 thus follows a so-called “ethics-by-design” approach: Ethical requirements and user perspectives are integrated into the technical development from the very beginning. The goal is to create digital technologies that are trustworthy, widely accepted, and practical for use in everyday clinical practice.