Bachelorarbeit
BA 450 für Bachelorstudierende (Betriebswirtschaft)
Allgemeines
| Verantwortlicher Dozent | Prof. Dr. Armin Heinzl |
|---|---|
| Veranstaltungsart | Thesis |
| Leistungspunkte | 12 ECTS |
| Sprache | Englisch |
| Infos für Studierende | Infos zu Bachelorarbeiten anderer Studiengänge finden Sie bei den Abschlussarbeiten |
Mechthild Pieper
Ansprechpartner Bachelorarbeit
Bei Fragen wenden Sie sich bitte an Mechthild Pieper.
Infos zur Veranstaltung
Ablauf
Deadline Event Thursday, 09.04.2026 Announcement of possible topics at 12pm / noon Wednesday 15.04.2026 Kick-off meeting and announcement of topic assignment Thursday 18.06.2026 Deadline for thesis submission until 12pm / noon Kurzbeschreibung
Individuals and organizations operate in a world that is increasingly permeated with digital technology. Every day we interact with Information Systems (IS) that make our phones smart, our cars safe, and our lives convenient. Likewise, Information Systems are embedded in the very core of the products, operations, and strategies of many organizations. Ever more, understanding and leveraging Information Systems is key to business success, not only for large and experienced players like SAP and Google, but also for small start-ups. The pervasive nature of digital technology is radically transforming our understanding of Information Systems, encompassing their development, coordination, use, and the way we interact with them. The primary objective of this seminar is to shed light on these issues and enrich our knowledge about how information systems impact organizations and individuals.
Themen
Topic Topic Description Supervisor AI-native social media platforms This bachelor thesis will focus on the dark side of AI-native social media platforms (e.g., Sora 2) and explore how AI-supported content and identity creation shape users’ experiences and behavior. The student’s core task is to develop a theoretically grounded empirical research design (preferably a quantitative experiment) that investigates a research question specific to this context, for example related to the Proteus Effect, self-objectification, or malicious manipulation of self-presentation with AI. The thesis should (1) identify a gap in extant literature, (2) motivate and define a clear research question, (3) derive hypotheses from an appropriate theoretical framework, (4) propose a concrete study design including measures and manipulations, and (5) if feasible, conduct a small pretest of the design to assess clarity and plausibility or to evaluate potential stimuli/ manipulation. Rosa Holtzwart AI-native social media platforms This bachelor thesis will focus on the dark side of AI-native social media platforms (e.g., Sora 2) and explore how AI-supported content and identity creation shape users’ experiences and behavior. The student’s core task is to develop a theoretically grounded empirical research design (preferably a quantitative experiment) that investigates a research question specific to this context, for example related to the Proteus Effect, self-objectification, or malicious manipulation of self-presentation with AI. The thesis should (1) identify a gap in extant literature, (2) motivate and define a clear research question, (3) derive hypotheses from an appropriate theoretical framework, (4) propose a concrete study design including measures and manipulations, and (5) if feasible, conduct a small pretest of the design to assess clarity and plausibility or to evaluate potential stimuli/ manipulation. Dr. Deborah Mateja Human-AI collaboration; sequence analysis; process mining In this bachelor thesis, you will conduct a sequence analysis (using R and/ or python and/ or process mining tools) of thousands of human-AI image generation interactions/ sessions. Our trace data includes all prompts and the generated images. The student will review relevant literature in the field of creativity, cognitive psychology, and collaboration literature to identify a suitable theory to apply to understand the sequence patterns identified during the sequence analysis. Successful completion will include understanding mechanisms that increase emergence, i.e., the space of possible action, in human-AI collaboration. Dr. Deborah Mateja Human-AI-Collaboration; Qualitative Data Analysis The rise of generative AI has transformed creative practices, enabling novel forms of human-AI collaboration in artistic image generation. Understanding how humans interact with AI systems in these processes is essential for both designing AI tools and gaining insights into emergent creativity. In this thesis, the student is expected to review and synthesize relevant literature on human-AI collaboration, identifying theoretical frameworks suitable for analyzing creative interactions and highlighting tensions or contradictions in collaborative practices. Building on this, the student will apply their insights to qualitatively analyze human-AI image generation interactions, including prompts, generated images, and associated quantitative measures. The thesis should ultimately provide a deeper understanding of how human-AI interactions shape creative processes and reveal factors that enhance or constrain emergent possibilities in co-creative workflows. Mechthild Pieper Healthcare, Hospital AI Platforms At the research campus M2OLIE at the University Hospital Mannheim, we are developing a medical-technology infrastructure that makes diagnosis and treatment of people with cancer possible within one day. As part of a prospective business model, the set-up on an AI platform is explored. In this context, this work explores hospital AI platforms as digital architectures through which hospitals integrate AI applications, data, and stakeholders across clinical and organizational processes. It aims at identifying underexplored areas in how such platforms are outlined in the literature, which technological and governance structures they require, and what opportunities and challenges they create for innovation, participation, and value creation in clinical settings. Michael Sternberg Healthcare, Clinical Business Model Innovation At the research campus M2OLIE, a clinic is currently under development, which will integrate the digital, medical, and processual innovations developed at the research campus, putting the substantially shortened diagnosis and treatment cycle of cancer patients into clinical practice. For this planned entity, a viable business plan and model are needed. This work sets out to explore what frontier developments exist in the realm of public clinical business models and if those (or elements of those) could be adopted. This work will include a mini case to determine the elements and challenges that need to be especially catered for in a business plan for the prospective M2OLIE-clinic. Michael Sternberg Healthcare, Technology & Patient centricity At the research campus M2OLIE, we are currently conducting a clinical trial to test our shortened diagnosis and treatment process, before translating it into the M2OLIE-clinic. In addition to the clinical dimensions relevant to the study, we want to incorporate the perspectives of patients. To do this, we will survey patients about their experiences with our process after (and before) they went through the trial. This Bachelor thesis aims to conceptualize and develop (not conduct!) a survey for these patients. Drawing from the literature, this work outlines an appropriate study format, and the dimensions and items necessary to validly measure the patient-centered performance of our process. Michael Sternberg Healthcare, Patient Journey Modelling To enable the diagnosis and treatment of cancer patients within one day, intimate knowledge of the respective care paths is required. The research will focus on the oncology patient journey, particularly in Germany or Europe, from initial suspicion through diagnosis and treatment decision-making. By employing a process-oriented Information Systems perspective, such as BPMN, the thesis will identify and model key data, including care stages, involved actors, their sequence, critical decision points, and associated probabilities. Building on previous seminar work, this study will significantly enhance the digital simulation of a future M2OLIE-clinic, contributing to more efficient and effective cancer care pathways. Michael Sternberg AI Avatars, HCI, Psychology, Self Disclosure This thesis investigates the phenomenon of self-disclosure of humans in interactions between humans and avatars, with a particular emphasis on applications beyond entertainment and in emerging everyday contexts like therapy avatars. The work will consist of a structured literature review grounded in Information Systems and Human-Computer Interaction. Students are expected to synthesize, map and critically evaluate existing research, identify gaps in the literature, and reflect on implications for the design of avatar-based systems. Particular attention should be paid to real-world applications (e.g., virtual agents, embodied AI), where theoretical insights from HCI are well developed but empirical evidence remains limited. Bastian Schieck AI Avatars, HCI, Sociology This thesis examines how humans interpret and respond to social roles based on visual and behavioral cues, and how these processes extend to interactions with avatars. The work will consist of a structured literature review within Information Systems and Human-Computer Interaction, engaging with concepts such as social categorization, role congruence, and anthropomorphism. Students will synthesize existing research on how role-related cues (e.g., appearance, behavior, contextual framing) influence perceptions of authority, competence, and trust in human interactions as a baseline and in avatar-based interactions. In addition, the thesis should explore how exposure to avatars in specific roles may shape or recalibrate users’ expectations and perceptions of real-world professionals, particularly in emerging application domains such as virtual agents and embodied AI. Bastian Schieck AI Avatars, Experimental Design, Research This thesis focuses on the development of experimental designs to investigate key phenomena in human–avatar interaction, specifically self-disclosure and role understanding. Building on existing literature in Information Systems and Human-Computer Interaction, the student will derive and propose four methodologically sound experimental setups, covering both quantitative and qualitative approaches for each phenomenon, grounded in prior work on avatar design and experimental research. The designs should incorporate advanced interaction conditions, including the ability for participants to create their own avatars and to experience them in life-sized representations. Students are expected to justify their design choices based on the literature, define procedures and measures, and critically reflect on methodological challenges and implications for evaluating avatar-based systems. Bastian Schieck Mobile App Development for Cancer Patients/ Patient-Centered Quality Management The systematic collection of Patient-Reported Outcome Measures (PROMs) and Patient-Reported Experience Measures (PREMs) is increasingly recognized as essential for patient-centered quality management in oncology. While validated instruments exist for long-term cancer care, their applicability to time-compressed, digitally supported treatment pathways like the M²OLIE Closed-Loop process for oligometastatic patients remains underexplored. Mobile patient applications offer new opportunities for real-time, low-burden data collection, yet selecting and integrating appropriate instruments raises questions around psychometric suitability, digital adaptability, patient burden, and interoperability with clinical data infrastructures. This bachelor thesis aims to conduct a structured literature review on PROMs and PREMs used in digital oncological settings, evaluate their suitability for short, intensive patient journeys, and develop a selection and integration framework for the M²OLIE mobile patient application. The objective is to identify which instruments best balance clinical validity, patient acceptability, and technical feasibility, and to propose a conceptual design for their implementation within the existing app architecture. Key starting points for research include: Kendir, C., Bousquet, J., Glennie, R. et al. The use of patient-reported outcome and experience measures for health policy purposes: A scoping review in oncology. Health Policy 127, 13–39 (2023). Bauer, J.S., Grawunder, S., Gensorowsky, D. et al. Use of Patient-Reported Outcome Measures and Patient-Reported Experience Measures Within Evaluation Studies of Telemedicine Applications: Systematic Review. J Med Internet Res 23(12), e30042 (2021). Farič, N., Scherrens, AL., Raemdonck, E. et al. Adaptation of digital integration of PROMs and PREMs in oncology during implementation: a scoping review. Support Care Cancer 34, 316 (2026). Florian Rüffer Explainable AI through Generative Models The explainability of AI decisions is a critical challenge in enhancing transparency and trust in AI systems. Traditional methods often fall short in providing clear, accessible explanations, particularly for complex models. Generative AI, including models like ChatGPT, has emerged as a promising solution by generating human-like explanations for AI decisions. However, this approach presents challenges related to the accuracy, reliability, and potential biases in AI-generated explanations, as well as the risk of oversimplification. This bachelor thesis aims to review the existing literature on the use of generative AI for explainability, focusing on how these models can generate explanations for AI decision-making processes. The objective is to evaluate the benefits and limitations of this approach and identify key challenges in ensuring that AI explanations are meaningful, accurate, and trustworthy. Key starting points for research include: Slack, D., Krishna, S., Lakkaraju, H. et al. Explaining machine learning models with interactive natural language conversations using TalkToModel. Nat Mach Intell 5, 873–883 (2023). https://doi.org/10.1038/s42256-023-00692-8 Schneider, J. Explainable Generative AI (GenXAI): a survey, conceptualization, and research agenda. Artif Intell Rev 57, 289 (2024). https://doi.org/10.1007/s10462-024-10916-x Florian Rüffer Theorizing Explainable AI The explainability of clinical AI systems remains a critical challenge for adoption in healthcare. Current dominant paradigms, additive attribution methods such as SHAP and LIME and counterfactual explanation methods, face well-documented limitations in clinical settings: attribution methods produce explanations clinicians frequently judge unhelpful, while counterfactual methods often propose unrealistic modifications in high-dimensional patient data. Meanwhile, clinicians naturally reason by comparing a patient to similar prior cases with known outcomes, yet this comparative reasoning style lacks a unified theoretical grounding in the explainable AI literature. This bachelor thesis aims to develop a theoretical framework for peer-relative explanation methods, in which predictions are explained by quantifying how a patient deviates from outcome-conditioned groups of similar patients. Drawing on exemplar-based categorization theory from cognitive science, the thesis will formalize the structural properties that distinguish comparative explanations from existing paradigms and derive criteria such as faithfulness, clinical plausibility, and meaningful null behavior that such explanations should satisfy. Key starting points for research include: Medin, D.L. & Schaffer, M.M. Context Theory of Classification Learning. Psychological Review 85(3), 207–238 (1978). Keane, M.T. & Smyth, B. Good Counterfactuals and Where to Find Them: A Case-Based Technique for Generating Counterfactuals for Explainable AI (XAI). ICCBR 2020. Florian Rüffer Digital platforms and regulation The rapid scaling and unparalleled market dominance of digital platforms have led to heightened attention from regulators worldwide. Europe has moved beyond traditional data protection towards a sophisticated “ex-ante” regulatory landscape that directly targets the structural foundations of platform business models. With the introduction of significant regulatory initiatives such as the Digital Markets Act (DMA), the Digital Services Act (DSA), the EU aims at ensuring fairness and contestability in digital markets. Most recently, the EU AI Act has introduced stringent requirements for data-driven business models and AI services embedded within these digital ecosystems. While these regulations offer significant benefits for users and market health, they may also trigger considerable changes in platform strategy. A bachelor thesis within this topic area aims to systematically synthesize extant research on regulation within digital platform contexts. Students will elaborate on pre-defined regulations and their impact on digital platform business models and identify dominant academic themes and research gaps. Dr. André Halckenhäußer Literatur
- The literature review should be done by students independently
- The review should include electronic literature sources offered by the University of Mannheim (Rechercheportal) as well as sources available on the internet
- Overviews of literature sources are available at the Mannheim University Library
