Modelling Adverse Events with the TOP Phenotyping Framework.

The detection and prevention of medication-related health risks, such as medication-associated adverse events (AEs), is a major challenge in patient care. A systematic review on the incidence and nature of in-hospital AEs found that 9.2% of hospitalised patients suffer an AE, and approximately 43% of these AEs are considered to be preventable. Adverse events can be identified using algorithms that operate on electronic medical records (EMRs) and research databases. Such algorithms normally consist of structured filter criteria and rules to identify individuals with certain phenotypic traits, thus are referred to as phenotype algorithms. Many attempts have been made to create tools that support the development of algorithms and their application to EMRs. However, there are still gaps in terms of functionalities of such tools, such as standardised representation of algorithms and complex Boolean and temporal logic. In this work, we focus on the AE delirium, an acute brain disorder affecting mental status and attention, thus not trivial to operationalise in EMR data. We use this AE as an example to demonstrate the modelling process in our ontology-based framework (TOP Framework) for modelling and executing phenotype algorithms. The resulting semantically modelled delirium phenotype algorithm is independent of data structure, query languages and other technical aspects, and can be run on a variety of source systems in different institutions.

Concept Graphs: A Novel Approach for Textual Analysis of Medical Documents.

The task of automatically analyzing the textual content of documents faces a number of challenges in general but even more so when dealing with the medical domain. Here, we can't normally rely on specifically pre-trained NLP models or even, due to data privacy reasons, (massive) amounts of training material to generate said models. We, therefore, propose a method that utilizes general-purpose basic text analysis components and state-of-the-art transformer models to represent a corpus of documents as multiple graphs, wherein important conceptually related phrases from documents constitute the nodes and their semantic relation form the edges. This method could serve as a basis for several explorative procedures and is able to draw on a plethora of publicly available resources. We test it by comparing the effectiveness of these so-called Concept Graphs with another recently suggested approach for a common use case in information retrieval, document clustering.

Problems in FAIRifying Medical Datasets.

Despite their young age, the FAIR principles are recognised as important guidelines for research data management. Their generic design, however, leaves much room for interpretation in domain-specific application. Based on practical experience in the operation of a data repository, this article addresses problems in FAIR provisioning of medical data for research purposes in the use case of the Leipzig Health Atlas project and shows necessary future developments.

3000PA-Towards a National Reference Corpus of German Clinical Language.

We introduce 3000PA, a clinical document corpus composed of 3,000 EPRs from three different clinical sites, which will serve as the backbone of a national reference language resource for German clinical NLP. We outline its design principles, results from a medication annotation campaign and the evaluation of a first medication information extraction prototype using a subset of 3000PA.

CDA-Compliant Section Annotation of German-Language Discharge Summaries: Guideline Development, Annotation Campaign, Section Classification.

We present the outcome of an annotation effort targeting the content-sensitive segmentation of German clinical reports into sections. We recruited an annotation team of up to eight medical students to annotate a clinical text corpus on a sentence-by-sentence basis in four pre-annotation iterations and one final main annotation step. The annotation scheme we came up with adheres to categories developed for clinical documents in the HL7-CDA (Clinical Document Architecture) standard for section headings. Once the scheme became stable, we ran the main annotation campaign on the complete set of roughly 1,000 clinical documents. Due to its reliance on the CDA standard, the annotation scheme allows the integration of legacy and newly produced clinical documents within a common pipeline. We then made direct use of the annotations by training a baseline classifier to automatically identify sections in clinical reports.

Scholarly Information Extraction Is Going to Make a Quantum Leap with PubMed Central (PMC).

With the increasing availability of complete full texts (journal articles), rather than their surrogates (titles, abstracts), as resources for text analytics, entirely new opportunities arise for information extraction and text mining from scholarly publications. Yet, we gathered evidence that a range of problems are encountered for full-text processing when biomedical text analytics simply reuse existing NLP pipelines which were developed on the basis of abstracts (rather than full texts). We conducted experiments with four different relation extraction engines all of which were top performers in previous BioNLP Event Extraction Challenges. We found that abstract-trained engines loose up to 6.6% F-score points when run on full-text data. Hence, the reuse of existing abstract-based NLP software in a full-text scenario is considered harmful because of heavy performance losses. Given the current lack of annotated full-text resources to train on, our study quantifies the price paid for this short cut.

Sharing models and tools for processing German clinical texts.

The automatic processing of non-English clinical documents is massively hampered by the lack of publicly available medical language resources for training, testing and evaluating NLP components. We suggest sharing statistical models derived from access-protected clinical documents as a reasonable substitute and provide solutions for sentence splitting, tokenization and POS tagging of German clinical texts. These three components were trained on the confidential FRAMED corpus, a non-sharable collection of various German-language clinical document types. The models derived therefrom outperform alternative components from OPENNLP and the Stanford POS tagger, also trained on FRAMED.