Learning Portuguese Clinical Word Embeddings: A Multi-Specialty and Multi-Institutional Corpus of Clinical Narratives Supporting a Downstream Biomedical Task.

In this paper, we trained a set of Portuguese clinical word embedding models of different granularities from multi-specialty and multi-institutional clinical narrative datasets. Then, we assessed their impact on a downstream biomedical NLP task of Urinary Tract Infection disease identification. Additionally, we intrinsically evaluated our main model using an adapted version of Bio-SimLex for the Portuguese language. Our empirical results showed that the larger, coarse-grained model achieved a slightly better outcome when compared with the small, fine-grained model in the proposed task. Moreover, we obtained satisfactory results with Bio-SimLex intrinsic evaluation.

A Domain Knowledge-Enhanced LSTM-CRF Model for Disease Named Entity Recognition.

Disease named entity recognition (NER) is a critical task for most biomedical natural language processing (NLP) applications. For example, extracting diseases from clinical trial text can be helpful for patient profiling and other downstream applications such as matching clinical trials to eligible patients. Similarly, disease annotation in biomedical articles can help information search engines to accurately index them such that clinicians can easily find relevant articles to enhance their knowledge. In this paper, we propose a domain knowledge-enhanced long short-term memory network-conditional random field (LSTM-CRF) model for disease named entity recognition, which also augments a character-level convolutional neural network (CNN) and a character-level LSTM network for input embedding. Experimental results on a scientific article dataset show the effectiveness of our proposed models compared to state-of-the-art methods in disease recognition.

Comparative effectiveness of convolutional neural network (CNN) and recurrent neural network (RNN) architectures for radiology text report classification.

This paper explores cutting-edge deep learning methods for information extraction from medical imaging free text reports at a multi-institutional scale and compares them to the state-of-the-art domain-specific rule-based system - PEFinder and traditional machine learning methods - SVM and Adaboost. We proposed two distinct deep learning models - (i) CNN Word - Glove, and (ii) Domain phrase attention-based hierarchical recurrent neural network (DPA-HNN), for synthesizing information on pulmonary emboli (PE) from over 7370 clinical thoracic computed tomography (CT) free-text radiology reports collected from four major healthcare centers. Our proposed DPA-HNN model encodes domain-dependent phrases into an attention mechanism and represents a radiology report through a hierarchical RNN structure composed of word-level, sentence-level and document-level representations. Experimental results suggest that the performance of the deep learning models that are trained on a single institutional dataset, are better than rule-based PEFinder on our multi-institutional test sets. The best F1 score for the presence of PE in an adult patient population was 0.99 (DPA-HNN) and for a pediatrics population was 0.99 (HNN) which shows that the deep learning models being trained on adult data, demonstrated generalizability to pediatrics population with comparable accuracy. Our work suggests feasibility of broader usage of neural network models in automated classification of multi-institutional imaging text reports for a variety of applications including evaluation of imaging utilization, imaging yield, clinical decision support tools, and as part of automated classification of large corpus for medical imaging deep learning work.

Translation of UMLS ontologies from european portuguese to brazilian portuguese / Translation of UMLS ontologies from european portuguese to brazilian portuguese

Ontologias terminológicas padronizadas e corretamente traduzidas são essenciais para o desenvolvimento de aplicações de processamento de linguagem natural na área da saúde. Para o desenvolvimento de uma aplicação de busca semântica em narrativas clínicas em português se fez necessária a utilização dos termos clínicos da Unified Medical Language System (UMLS). OBJETIVOS: Traduzir termos da UMLS em Português Europeu para Português Brasileiro. MÉTODOS: Foi desenvolvido um algoritmo de tradução semi-automática baseada em regras de substituição de texto. RESULTADOS: Após execução do algoritmo e avaliação por parte de especialistas, o algoritmo deixou de traduzir corretamente apenas 0.1% dos termos da base de testes. CONCLUSÃO: A utilização do método proposto se mostrou efetivo na tradução dos termos da UMLS e pode auxiliar em posteriores adaptações de listagens em Português Europeu para Português Brasileiro.

Correctly translated and standardized clinical ontologies are essential for development of Natural LanguageProcessing application for the medical domain. To develop an ontology-driven semantic search application for Portuguese clinical notes we needed to implement the Unified Medical Language System (UMLS) ontologies, specifically for Brazilian Portuguese. OBJECTIVES: To translate UMLS terms from European Portuguese to Brazilian Portuguese. METHODS: To develop a semi-automatic translation algorithm based on string replacement rules. RESULTS: Following the experiments and specialists' evaluation the algorithm mis-translated only 0.1% of terms in our test set. CONCLUSION: The proposed method proved to be effective for UMLS clinical terms translation and can be useful for posterior adaption ofa set of clinical terms from European Portuguese to Brazilian Portuguese.

Ontology-Driven Semantic Search for Brazilian Portuguese Clinical Notes.

The emerging penetration of Health IT in Latin America (especially in Brazil) has exacerbated the ever-increasing amount of Electronic Health Record (EHR) clinical free text documents.This imposes a workflow efficiency challenge on clinicians who need to synthesize such documents during the typically time-constrained patient care. We propose an ontology-driven semantic search framework that effectively supports clinicians' information synthesis at the point of care.