-
Language corpora for the Dutch medical domain
Authors:
B. van Es
Abstract:
Background: Dutch medical corpora are scarce, limiting NLP development. Methods: We translated English datasets, identified medical text in generic corpora, and extracted open Dutch medical resources. Results: The resulting corpus comprises +- 36 billion tokens across the medical domain in about 105 million documents, freely available on Hugging Face. Conclusion: This work establishes the first la…
▽ More
Background: Dutch medical corpora are scarce, limiting NLP development. Methods: We translated English datasets, identified medical text in generic corpora, and extracted open Dutch medical resources. Results: The resulting corpus comprises +- 36 billion tokens across the medical domain in about 105 million documents, freely available on Hugging Face. Conclusion: This work establishes the first large-scale Dutch medical language corpus for pre-training and downstream NLP tasks.
△ Less
Submitted 11 August, 2026; v1 submitted 28 April, 2026;
originally announced April 2026.
-
Automatic Cardiac Risk Management Classification using large-context Electronic Patients Health Records
Authors:
Jacopo Vitale,
David Della Morte,
Luca Bacco,
Mario Merone,
Mark de Groot,
Saskia Haitjema,
Leandro Pecchia,
Bram van Es
Abstract:
To overcome the limitations of manual administrative coding in geriatric Cardiovascular Risk Management, this study introduces an automated classification framework leveraging unstructured Electronic Health Records (EHRs). Using a dataset of 3,482 patients, we benchmarked three distinct modeling paradigms on longitudinal Dutch clinical narratives: classical machine learning baselines, specialized…
▽ More
To overcome the limitations of manual administrative coding in geriatric Cardiovascular Risk Management, this study introduces an automated classification framework leveraging unstructured Electronic Health Records (EHRs). Using a dataset of 3,482 patients, we benchmarked three distinct modeling paradigms on longitudinal Dutch clinical narratives: classical machine learning baselines, specialized deep learning architectures optimized for large-context sequences, and general-purpose generative Large Language Models (LLMs) in a zero-shot setting. Additionally, we evaluated a late fusion strategy to integrate unstructured text with structured medication embeddings and anthropometric data. Our analysis reveals that the custom Transformer architecture outperforms both traditional methods and generative \acs{llm}s, achieving the highest F1-scores and Matthews Correlation Coefficients. These findings underscore the critical role of specialized hierarchical attention mechanisms in capturing long-range dependencies within medical texts, presenting a robust, automated alternative to manual workflows for clinical risk stratification.
△ Less
Submitted 10 March, 2026;
originally announced March 2026.
-
Tackling the Algorithmic Control Crisis -- the Technical, Legal, and Ethical Challenges of Research into Algorithmic Agents
Authors:
B. Bodo,
N. Helberger,
K. Irion,
F. Zuiderveen Borgesius,
J. Moller,
B. Van der Velde,
N. Bol,
B. van Es,
C. de Vreese
Abstract:
Algorithmic agents permeate every instant of our online existence. Based on our digital profiles built from the massive surveillance of our digital existence, algorithmic agents rank search results, filter our emails, hide and show news items on social networks feeds, try to guess what products we might buy next for ourselves and for others, what movies we want to watch, and when we might be pregn…
▽ More
Algorithmic agents permeate every instant of our online existence. Based on our digital profiles built from the massive surveillance of our digital existence, algorithmic agents rank search results, filter our emails, hide and show news items on social networks feeds, try to guess what products we might buy next for ourselves and for others, what movies we want to watch, and when we might be pregnant. Algorithmic agents select, filter, and recommend products, information, and people. Increasingly, algorithmic agents don't just select from the range of human created alternatives, but also they create. Burgeoning algorithmic agents are capable of providing us with content made just for us, and engage with us through one-of-a-kind, personalized interactions. Studying these algorithmic agents presents a host of methodological, ethical, and logistical challenges. The objectives of our paper are two-fold. The first aim is to describe one possible approach to researching the individual and societal effects of algorithmic recommenders, and to share our experiences with the academic community. The second is to contribute to a more fundamental discussion about the ethical and legal issues of "tracking the trackers", as well as the costs and trade-offs involved. Our paper will contribute to the discussion on the relative merits, costs and benefits of different approaches to ethically and legally sound research on algorithmic governance. We will argue that besides shedding light on how users interact with algorithmic agents, we also need to be able to understand how different methods of monitoring our algorithmically controlled digital environments compare to each other in terms of costs and benefits. We conclude our article with a number of concrete suggestions for how to address the practical, ethical and legal challenges of researching algorithms and their effects on users and society.
△ Less
Submitted 29 October, 2025;
originally announced October 2025.
-
Diagnosis extraction from unstructured Dutch echocardiogram reports using span- and document-level characteristic classification
Authors:
Bauke Arends,
Melle Vessies,
Dirk van Osch,
Arco Teske,
Pim van der Harst,
René van Es,
Bram van Es
Abstract:
Clinical machine learning research and AI driven clinical decision support models rely on clinically accurate labels. Manually extracting these labels with the help of clinical specialists is often time-consuming and expensive. This study tests the feasibility of automatic span- and document-level diagnosis extraction from unstructured Dutch echocardiogram reports. We included 115,692 unstructured…
▽ More
Clinical machine learning research and AI driven clinical decision support models rely on clinically accurate labels. Manually extracting these labels with the help of clinical specialists is often time-consuming and expensive. This study tests the feasibility of automatic span- and document-level diagnosis extraction from unstructured Dutch echocardiogram reports. We included 115,692 unstructured echocardiogram reports from the UMCU a large university hospital in the Netherlands. A randomly selected subset was manually annotated for the occurrence and severity of eleven commonly described cardiac characteristics. We developed and tested several automatic labelling techniques at both span and document levels, using weighted and macro F1-score, precision, and recall for performance evaluation. We compared the performance of span labelling against document labelling methods, which included both direct document classifiers and indirect document classifiers that rely on span classification results. The SpanCategorizer and MedRoBERTa$.$nl models outperformed all other span and document classifiers, respectively. The weighted F1-score varied between characteristics, ranging from 0.60 to 0.93 in SpanCategorizer and 0.96 to 0.98 in MedRoBERTa$.$nl. Direct document classification was superior to indirect document classification using span classifiers. SetFit achieved competitive document classification performance using only 10% of the training data. Utilizing a reduced label set yielded near-perfect document classification results. We recommend using our published SpanCategorizer and MedRoBERTa$.$nl models for span- and document-level diagnosis extraction from Dutch echocardiography reports. For settings with limited training data, SetFit may be a promising alternative for document classification.
△ Less
Submitted 15 August, 2024; v1 submitted 13 August, 2024;
originally announced August 2024.
-
Negation detection in Dutch clinical texts: an evaluation of rule-based and machine learning methods
Authors:
Bram van Es,
Leon C. Reteig,
Sander C. Tan,
Marijn Schraagen,
Myrthe M. Hemker,
Sebastiaan R. S. Arends,
Miguel A. R. Rios,
Saskia Haitjema
Abstract:
As structured data are often insufficient, labels need to be extracted from free text in electronic health records when developing models for clinical information retrieval and decision support systems. One of the most important contextual properties in clinical text is negation, which indicates the absence of findings. We aimed to improve large scale extraction of labels by comparing three method…
▽ More
As structured data are often insufficient, labels need to be extracted from free text in electronic health records when developing models for clinical information retrieval and decision support systems. One of the most important contextual properties in clinical text is negation, which indicates the absence of findings. We aimed to improve large scale extraction of labels by comparing three methods for negation detection in Dutch clinical notes. We used the Erasmus Medical Center Dutch Clinical Corpus to compare a rule-based method based on ContextD, a biLSTM model using MedCAT and (finetuned) RoBERTa-based models. We found that both the biLSTM and RoBERTa models consistently outperform the rule-based model in terms of F1 score, precision and recall. In addition, we systematically categorized the classification errors for each model, which can be used to further improve model performance in particular applications. Combining the three models naively was not beneficial in terms of performance. We conclude that the biLSTM and RoBERTa-based models in particular are highly accurate accurate in detecting clinical negations, but that ultimately all three approaches can be viable depending on the use case at hand.
△ Less
Submitted 1 September, 2022;
originally announced September 2022.