Medizinische Universität Graz Austria/Österreich - Forschungsportal - Medical University of Graz
Gewählte Publikation:
SHR Neuro Krebs Kardio Lipid Stoffw Microb
Kraisnikovic, C; Harb, R; Plass, M; Al Zoughbi, W; Holzinger, A; Müller, H.
Fine-tuning language model embeddings to reveal domain knowledge: An explainable artificial intelligence perspective on medical decision making
ENG APPL ARTIF INTEL. 2025; 139: 109561
Doi: 10.1016/j.engappai.2024.109561
Web of Science
FullText
FullText_MUG
- Führende Autor*innen der Med Uni Graz
- Kraisnikovic Ceca
- Plass Markus
- Co-Autor*innen der Med Uni Graz
- Al-Zoughbi Wael
- Harb Robert
- Holzinger Andreas
- Müller Heimo
- Altmetrics:
- Dimensions Citations:
- Plum Analytics:
- Scite (citation analytics):
- Abstract:
- Integrating large language models (LLMs) to retrieve targeted medical knowledge from electronic health records enables significant advancements in medical research. However, recognizing the challenges associated with using LLMs in healthcare is essential for successful implementation. One challenge is that medical records combine unstructured textual information with highly sensitive personal data. This, in turn, highlights the need for explainable Artificial Intelligence (XAI) methods to understand better how LLMs function in the medical domain. In this study, we propose a novel XAI tool to accelerate data-driven cancer research. We apply the Bidirectional Encoder Representations from Transformers (BERT) model to German language pathology reports examining the effects of domain-specific language adaptation and fine-tuning. We demonstrate our model on a real-world pathology dataset, analyzing the contextual representations of diagnostic reports. By illustrating decisions made by fine-tuned models, we provide decision values that can be applied in medical research. To address interpretability, we conduct a performance evaluation of the classifications generated by our fine-tuned model, as assessed by an expert pathologist. In domains such as medicine, inspection of the medical knowledge map in conjunction with expert evaluation reveals valuable information about how contextual representations of key disease features are categorized. This ultimately benefits data structuring and labeling and paves the way for even more advanced approaches to XAI, combining text with other input modalities, such as images which are then applicable to various engineering problems.
- Find related publications in this database (Keywords)
- Pathology reports
- Large language models in pathology
- Bidirectional Encoder Representations from Transformers model
- Language model for German
- Domain-language adaptation
- Fine-tuning
- Analysis of embeddings
- Pathology-specific tasks
- Digital pathology
- Interpretable medical decision scores