Journal of Biomedical Semantics最新文献

Background: The high volume of research focusing on extracting patient information from electronic health records (EHRs) has led to an increase in the demand for annotated corpora, which are a precious resource for both the development and evaluation of natural language processing (NLP) algorithms. The absence of a multipurpose clinical corpus outside the scope of the English language, especially in Brazilian Portuguese, is glaring and severely impacts scientific progress in the biomedical NLP field.

Methods: In this study, a semantically annotated corpus was developed using clinical text from multiple medical specialties, document types, and institutions. In addition, we present, (1) a survey listing common aspects, differences, and lessons learned from previous research, (2) a fine-grained annotation schema that can be replicated to guide other annotation initiatives, (3) a web-based annotation tool focusing on an annotation suggestion feature, and (4) both intrinsic and extrinsic evaluation of the annotations.

Results: This study resulted in SemClinBr, a corpus that has 1000 clinical notes, labeled with 65,117 entities and 11,263 relations. In addition, both negation cues and medical abbreviation dictionaries were generated from the annotations. The average annotator agreement score varied from 0.71 (applying strict match) to 0.92 (considering a relaxed match) while accepting partial overlaps and hierarchically related semantic types. The extrinsic evaluation, when applying the corpus to two downstream NLP tasks, demonstrated the reliability and usefulness of annotations, with the systems achieving results that were consistent with the agreement scores.

Conclusion: The SemClinBr corpus and other resources produced in this work can support clinical NLP studies, providing a common development and evaluation resource for the research community, boosting the utilization of EHRs in both clinical practice and biomedical research. To the best of our knowledge, SemClinBr is the first available Portuguese clinical corpus.

Background: Transfer learning is a common practice in image classification with deep learning where the available data is often limited for training a complex model with millions of parameters. However, transferring language models requires special attention since cross-domain vocabularies (e.g. between two different modalities MR and US) do not always overlap as the pixel intensity range overlaps mostly for images.

Method: We present a concept of similar domain adaptation where we transfer inter-institutional language models (context-dependent and context-independent) between two different modalities (ultrasound and MRI) to capture liver abnormalities.

Results: We use MR and US screening exam reports for hepatocellular carcinoma as the use-case and apply the transfer language space strategy to automatically label imaging exams with and without structured template with > 0.9 average f1-score.

Conclusion: We conclude that transfer learning along with fine-tuning the discriminative model is often more effective for performing shared targeted tasks than the training for a language space from scratch.

Background: Population-based cancer registries are a critical reference source for the surveillance and control of cancer. Cancer registries work extensively with the internationally recognised TNM classification system used to stage solid tumours, but the system is complex and compounded by the different TNM editions in concurrent use. TNM ontologies exist but the design requirements are different for the needs of the clinical and cancer-registry domains. Two TNM ontologies developed specifically for cancer registries were designed for different purposes and have limitations for serving wider application. A unified ontology is proposed to serve the various cancer registry TNM-related tasks and reduce the multiplication effects of different ontologies serving specific tasks. The ontology is comprehensive of the rules for TNM edition 7 as required by cancer registries and designed on a modular basis to allow extension to other TNM editions.

Results: A unified ontology was developed building on the experience and design of the existing ontologies. It follows a modular approach allowing plug in of components dependent upon any particular TNM edition. A Java front-end was developed to interface with the ontology via the Web Ontology Language application programme interface and enables batch validation or classification of cancer registry records. The programme also allows the means of automated error correction in some instances. Initial tests verified the design concept by correctly inferring TNM stage and successfully handling the TNM-related validation checks on a number of cancer case records, with a performance similar to that of an existing ontology dedicated to the task.

Conclusions: The unified ontology provides a multi-purpose tool for TNM-related tasks in a cancer registry and is scalable for different editions of TNM. It offers a convenient way of quickly checking validity of cancer case stage information and for batch processing of multi-record data via a dedicated front-end programme. The ontology is adaptable to many uses, either as a standalone TNM module or as a component in applications of wider focus. It provides a first step towards a single, unified TNM ontology for cancer registries.

Background: Artificial intelligence methods applied to electronic medical records (EMRs) hold the potential to help physicians save time by sharpening their analysis and decisions, thereby improving the health of patients. On the one hand, machine learning algorithms have proven their effectiveness in extracting information and exploiting knowledge extracted from data. On the other hand, knowledge graphs capture human knowledge by relying on conceptual schemas and formalization and supporting reasoning. Leveraging knowledge graphs that are legion in the medical field, it is possible to pre-process and enrich data representation used by machine learning algorithms. Medical data standardization is an opportunity to jointly exploit the richness of knowledge graphs and the capabilities of machine learning algorithms.

Methods: We propose to address the problem of hospitalization prediction for patients with an approach that enriches vector representation of EMRs with information extracted from different knowledge graphs before learning and predicting. In addition, we performed an automatic selection of features resulting from knowledge graphs to distinguish noisy ones from those that can benefit the decision making. We report the results of our experiments on the PRIMEGE PACA database that contains more than 600,000 consultations carried out by 17 general practitioners (GPs).

Results: A statistical evaluation shows that our proposed approach improves hospitalization prediction. More precisely, injecting features extracted from cross-domain knowledge graphs in the vector representation of EMRs given as input to the prediction algorithm significantly increases the F1 score of the prediction.

Conclusions: By injecting knowledge from recognized reference sources into the representation of EMRs, it is possible to significantly improve the prediction of medical events. Future work would be to evaluate the impact of a feature selection step coupled with a combination of features extracted from several knowledge graphs. A possible avenue is to study more hierarchical levels and properties related to concepts, as well as to integrate more semantic annotators to exploit unstructured data.

Background: Text mining can be applied to automate knowledge extraction from unstructured data included in medical reports and generate quality indicators applicable for medical documentation. The primary objective of this study was to apply text mining methodology for the analysis of polysomnographic medical reports in order to quantify sources of variation - here the diagnostic precision vs. the inter-rater variability - in the work-up of sleep-disordered breathing. The secondary objective was to assess the impact of a text block standardization on the diagnostic precision of polysomnography reports in an independent test set.

Results: Polysomnography reports of 243 laboratory-based overnight sleep investigations scored by 9 trained sleep specialists of the Sleep Center St. Gallen were analyzed using a text-mining methodology. Patterns in the usage of discriminating terms allowed for the characterization of type and severity of disease and inter-rater homogeneity. The variation introduced by the inter-rater (technician/physician) heterogeneity was found to be twice as high compared to the variation introduced by effective diagnostic information. A simple text block standardization could significantly reduce the inter-rater variability by 44%, enhance the predictive value and ultimately improve the diagnostic accuracy of polysomnography reports.

Conclusions: Text mining was successfully used to assess and optimize the quality, as well as the precision and homogeneity of medical reporting of diagnostic procedures - here exemplified with sleep studies. Text mining methodology could lay the ground for objective and systematic qualitative assessment of medical reports.

Background: Drug repurposing can improve the return of investment as it finds new uses for existing drugs. Literature-based analyses exploit factual knowledge on drugs and diseases, e.g. from databases, and combine it with information from scholarly publications. Here we report the use of the Open Discovery Process on scientific literature to identify non-explicit ties between a disease, namely epilepsy, and known drugs, making full use of available epilepsy-specific ontologies.

Results: We identified characteristics of epilepsy-specific ontologies to create subsets of documents from the literature; from these subsets we generated ranked lists of co-occurring neurological drug names with varying specificity. From these ranked lists, we observed a high intersection regarding reference lists of pharmaceutical compounds recommended for the treatment of epilepsy. Furthermore, we performed a drug set enrichment analysis, i.e. a novel scoring function using an adaptive tuning parameter and comparing top-k ranked lists taking into account the varying length and the current position in the list. We also provide an overview of the pharmaceutical space in the context of epilepsy, including a final combined ranked list of more than 70 drug names.

Conclusions: Biomedical ontologies are a rich resource that can be combined with text mining for the identification of drug names for drug repurposing in the domain of epilepsy. The ranking of the drug names related to epilepsy provides benefits to patients and to researchers as it enables a quick evaluation of statistical evidence hidden in the scientific literature, useful to validate approaches in the drug discovery process.