Bridging policy and technology: AI transparency in healthcare

Abstract

The healthcare sector faces growing pressure due to personnel shortages. Artificial intelligence (AI) could help to maintain healthcare accessible, affordable and of high quality. This thesis explores how AI-assisted screening can support creating a literature overview on transparency requirements in healthcare AI„focusing on how transparency influences trust, accountability and ethical deployment, and how such methods can be applied within governments or healthcare organizations. Two screening pipelines using active learning were evaluated to accelerate document selection. Pipeline 1 combines vector-based semantic search with active learning to screen scientific literature on AI transparency, using OpenAlex and ASReview. Pipeline 2 involves screening multilingual policy documents with ASReview. Both pipelines effectively identified relevant documents while reducing manual screening workload. The results demonstrate that AI-assisted screening can be applied responsibly in public institutions. With appropriate investment in preparation and user support, these methods can enhance the efficiency and quality of evidence-based policymaking.