Stefan Hegselmann

I am interested in machine learning for healthcare. My research focuses on using large generative models to enable efficient, high-performing predictions, and on applying natural language processing techniques to clinical text. I am particularly interested in using state-of-the-art machine learning methods to address meaningful medical problems and improve healthcare delivery.

I am currently an affiliated postdoctoral researcher at the Eils lab focussing on artificial intelligence in healthcare. I am also a resident physician at the Department of Cardiology, Angiology and Intensive Care Medicine at the Charité in Berlin, where I am involved in clinical practice. My goal is to bridge the gap between modern machine learning and clinical practice.

Previously, I have obtained a Master’s degree in computer science from RWTH Aachen University with stays at the University of Gothenburg and UC Berkeley. After that I obtained a medical degree (MD) and a PhD in computer science from the University of Münster and worked with the Clinical ML group at MIT.

You can find all my papers on my Google Scholar profile.

news

Apr 01, 2025	Transferred to this new website.

selected publications

Large language models are few-shot clinical information extractors

Monica Agrawal, Stefan Hegselmann, Hunter Lang, and 2 more authors

In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, Dec 2022

Abs DOI HTML

A long-running goal of the clinical NLP community is the extraction of important variables trapped in clinical notes. However, roadblocks have included dataset shift from the general domain and a lack of public clinical corpora and annotations. In this work, we show that large language models, such as InstructGPT (Ouyang et al., 2022), perform well at zero- and few-shot information extraction from clinical text despite not being trained specifically for the clinical domain. Whereas text classification and generation performance have already been studied extensively in such models, here we additionally demonstrate how to leverage them to tackle a diverse set of NLP tasks which require more structured outputs, including span identification, token-level sequence classification, and relation extraction. Further, due to the dearth of available data to evaluate these systems, we introduce new datasets for benchmarking few-shot clinical information extraction based on a manual re-annotation of the CASI dataset (Moon et al., 2014) for new tasks. On the clinical extraction tasks we studied, the GPT-3 systems significantly outperform existing zero- and few-shot baselines.
Tabllm: Few-shot classification of tabular data with large language models

Stefan Hegselmann, Alejandro Buendia, Hunter Lang, and 3 more authors

In International Conference on Artificial Intelligence and Statistics, Dec 2023

HTML
A data-centric approach to generate faithful and high quality patient summaries with large language models

Stefan Hegselmann, Shannon Zejiang Shen, Florian Gierse, and 3 more authors

Conference on Health, Inference, and Learning 2024, Dec 2024

HTML