Working Groups

Generative AI and Personalization

The “Generative AI and Personalization” working group investigates how generative AI models can advance personalized learning in simulation-based learning environments. Taking current developments into account, it explores new ways of adapting, creating, and delivering content tailored to individual learners. These range from the dynamic adaptation of simulations to scalable instructional support, such as customized feedback. The group also examines how generative AI can create diverse, multimodal learning materials that reflect authentic professional practice. In addition to large language models, its members consider smaller, openly accessible models that can be fine-tuned for specific purposes, as well as language models used as middleware to coordinate multiple AI systems. Across all SHARP projects within CRC 419, the group supports the selection of suitable AI models for personalization, taking into account the authenticity of the respective learning environment, the degree of personalization, and the mode of interaction.
The Working Group is led by Prof. Dr Stefan Küchemann.

Transfer

The CRC SHARP can draw on the strong support of both LMU’s Research and Technology Transfer Unit, the joint Technology Transfer Office (TTO) of LMU and its University Hospital, and TUM’s Office for Research and Innovation (TUM ForTe). One key challenge in SHARP will be to transfer the research results and insights to university teachers responsible for teacher and medical education at the universities participating in the TRR, as well as at the national level. Therefore, the transfer strategy includes establishing a training program for higher education teachers to use adaptive simulations in their courses and lectures effectively. Another part of the transfer strategy is to identify and contact stakeholders such as deans of study and coordination boards, as well as professional associations, for an exchange on the research and the findings of the SHARP projects.
The Working Group is led by Prof. Dr Bernd Schmidt-Herta.

Open Science

The Open Science Working Group encourages and supports the implementation of Open Science best practices and tools to ensure that our research is FAIR, transparent, and reusable. We support researchers throughout the entire research lifecycle, from study planning and preregistration to research data management, reproducible analyses, FAIR data sharing, and open access publishing.
To achieve this, we develop consortium-wide standards and practical resources while also addressing researchers’ needs through workshops, training sessions, and curated materials, often in collaboration with the LMU Open Science Centerand other partners.
We also promote innovative approaches, such as agile research data management or “Repro-Games”, to strengthen the quality, robustness, and reproducibility of our research. Our ultimate goal is to embed Open Science in everyday research practice while empowering SHARP researchers to become responsible, transparent, and collaborative scientists who contribute to high-quality, cumulative research.
The Working Group is led by Prof. Dr Mario Gollwitzer and Dr Markus Gödker.

Science Communication

The Science Communication Working Group strengthens SHARP’s external communication and contributes to a consistent and visible presence of the Collaborative Research Centre/Transregio. Its tasks include maintaining and further developing the SHARP website, publishing news, and communicating important updates and developments within the project. The Working Group also supports SHARP’s LinkedIn communication. In addition, it develops key materials and standards for external communication, such as templates and shared design principles. The Working Group is led by Prof. Dr Michael Sailer and Dr Nadine Hammele.

Gender & Diversity

The SHARP Gender & Diversity Working Group is committed to fostering an inclusive research environment that reflects the full diversity of the scientific community across dimensions such as gender, age, cultural and social background, sexual orientation, disability, and international experience. In close collaboration with the Gender and Diversity Consulting Services at LMU and TUM, the Working Group offers workshops on unconscious bias, inclusive leadership, mental health, and resilience. These activities are complemented by individual counselling and mentoring for structurally disadvantaged researchers at all career stages. SHARP already considerably exceeds national benchmarks for the representation of women in leadership positions: 38% of its Principal Investigators are women, compared with a national average of 31%. The Working Group is also advancing measures to achieve even greater gender parity in the coming years. Practical support includes a dedicated postdoctoral seed grant, a call for Equal Opportunities & Diversity funding, additional grants, and family-friendly event formats with hybrid participation options. The working group is led by Prof. Dr. Sarah Hofer as the elected Gender and Diversity Representative, supported by Deisy Montenegro and Dora Burbank as her deputies. Together, they advocate for gender equality and diversity within the consortium and ensure close coordination with the broader equal opportunities and diversity initiatives at LMU and TUM.

Learning Analytics for Personalisation Using Text Entries and Log Files

The Working Group addresses the methodological challenge of deriving actionable indicators for personalisation from unstructured process data, specifically free-text entries and interaction log files generated in simulation-based learning environments. The group is coordinated by Prof. Dr. Matthias Stadler and Dr. Constanze Richters (Institute of Medical Education, LMU University Hospital) and meets quarterly, in line with the CRC's methodological working group format. It primarily contributes to methodological topics (a) process analytics with ML and learning analytics for micro-adaptation and (c) NLP approaches to analysing natural language input and learner interaction data in simulation-based learning environments. The group's work currently builds on two complementary strands. The first strand focuses on the automated analysis of free-text student submissions (e.g., prescriptions, case documentation, justifications) using large language models to generate guideline-informed, individualised feedback at scale, addressing the persistent bottleneck of time-intensive and inconsistent human grading in large cohorts. The second strand focuses on the analysis of interaction log files from agent-based simulations, applying machine learning classification approaches (support vector machines, random forest, gradient boosting) to collaborative diagnostic activities in order to predict diagnostic accuracy and to identify the point in the diagnostic process at which reliable prediction becomes possible, providing a methodological basis for adaptive and meso-adaptive scaffolding. Across both strands, Working Group B pursues the following goals: (1) establishing methodological synergies across SHARP subprojects that work with text or log data, in coordination with Project INF; (2) conducting a systematic needs assessment among doctoral candidates to inform the design of project academies and workshops; and (3) providing a recurring forum in which methodological approaches, code, and preliminary results can be shared and discussed across project boundaries. Consistent with SHARP's open science commitments, the group encourages pre-registration of hypothesis-driven analyses, transparent reporting of feature extraction and model selection procedures, and the sharing of analysis pipelines to support reproducibility and reuse by other projects.