PhD position - Large Language Models & natural language processing to unlock the potential of healthcare

Are you enthusiastic about developing, evaluating and applying natural language processing and large language models for Dutch healthcare? Then continue reading!

Dit ga je doen

The Methods of Epidemiological Research team at the Julius Center of the UMC Utrecht is seeking an enthusiastic and ambitious PhD candidate. In this project, you will focus on developing, validating, and applying fine-grained natural language processing (NLP) and large language model (LLM) methods to unlock information from Dutch electronic health record (EHR) free text for secondary use in research.

Electronic health records contain a wealth of relevant patient information in unstructured free-text notes. While structured (coded) fields are commonly reused for research, a substantial amount of nuanced and context-rich patient information remains locked in the large amount of narrative texts in EHRs.

NLP and LLM-based methods offer great promise to unlock this important patient information, e.g. for secondary use in (bio)medical and epidemiological research. However, current methods and approaches typically map free text to structured data using rule-based methods, which may lead to inaccurate classifications. In addition, data for secondary use is currently mapped to coarse-grained ontologies or common data models (e.g., the Observational Medical Outcomes Partnership Common Data Model), which may lead to substantial information loss. Moreover, most medical NLP/LLM tools are developed and validated in English, leaving a major gap for Dutch EHR data. This PhD project aims to address these challenges by developing and validating fine-grained information extraction approaches for Dutch EHR texts, minimizing information loss while ensuring robustness, transparency, and practical usability.

You will work at the intersection of epidemiology, clinical research, data science, and AI, with a strong methodological focus.

Hier ga je werken

In this position, you will work and be supervised in the Department of Epidemiology and Health Economics as part of Methods of Epidemiological Research program, in the Julius Center at the UMC Utrecht, in close collaboration with the UMCU AI labs. You will be part of an energetic, enthusiastic team of more than 35 colleagues from very different backgrounds. The Julius Center has an extensive national and international network.

Dit neem je mee

We are looking for a candidate who:

• Holds an MSc degree in artificial intelligence, natural language processing, computer science, biostatistics, data science, (clinical) epidemiology, biomedical sciences, or a related field
• Has strong interest in methodological research in healthcare
• Has experience with NLP and/or language modeling techniques
• Has solid programming skills (e.g., Python)
• Is motivated to work at the interface of AI and clinical research
• Enjoys interdisciplinary collaboration
• Has excellent written and spoken English skills
• Proficiency in Dutch is a pre-requisite given the focus on unlocking Dutch EHR texts
• Experience with (bio)medical research, ontologies, semantic parsing, or privacy-preserving AI methods is considered a plus.

A.M.Leeuwenberg-15@umcutrecht.nl