Thema der Dissertation:
Human-in-the-loop and Storytelling-based Explainable Frameworks for the Next Generation of Explainable
Ensemble Methods and Artificial Intelligence Techniques Thema der Disputation:
Human-Centered Trustworthy Explainable Artificial Intelligence (XAI) in Health-Related Domains
Human-in-the-loop and Storytelling-based Explainable Frameworks for the Next Generation of Explainable
Ensemble Methods and Artificial Intelligence Techniques Thema der Disputation:
Human-Centered Trustworthy Explainable Artificial Intelligence (XAI) in Health-Related Domains
Abstract: Currently, the adoption of artificial intelligence (AI) in health-related domains remains largely confined to research settings. To enhance the adoption of AI in real-world health-related settings, the AI requires extensive validation and evaluation. While practitioners increasingly couple Explainable Artificial Intelligence (XAI) frameworks with AI and emphasize on the trustworthy AI, XAI is frequently implemented only as a post-hoc analysis tool. This research presents a different perspective, emphasizing that XAI must transition from a retrospective afterthought to a foundational, integrated stage of the AI development workflow. The presentation will first establish an important & crucial role of XAI in health-related domains and how to integrate it. It will then demonstrate how benchmark frameworks, SHapley Additive exPlanations (SHAP), must be coupled with reliability metrics, extending these principles across both tabular data and medical imaging. Finally, the research highlights the necessity of domain-centric explainability, showcasing how methodologies can be tailored for highly specific tasks, such as peptide analysis utilizing Protein Language Models (PLMs).
Zeit & Ort
07.07.2026 | 10:00
Seminarraum K40
(Fachbereich Mathematik und Informatik, Takustr. 9, 14195 Berlin)
&
WebEx