Building a specialized lexicon for breast cancer clinical trial subject eligibility analysis

Abstract

A natural language processing (NLP) application requires sophisticated lexical resources to support its processing goals. Different solutions, such as dictionary lookup and MetaMap, have been proposed in the healthcare informatics literature to identify disease terms with more than one word (multi-gram disease named entities). Although a lot of work has been done in the identification of protein- and gene-named entities in the biomedical field, not much research has been done on the recognition and resolution of terminologies in the clinical trial subject eligibility analysis. In this study, we develop a specialized lexicon for improving NLP and text mining analysis in the breast cancer domain, and evaluate it by comparing it with the Systematized Nomenclature of Medicine Clinical Terms (SNOMED CT). We use a hybrid methodology, which combines the knowledge of domain experts, terms from multiple online dictionaries, and the mining of text from sample clinical trials. Use of our methodology introduces 4243 unique lexicon items, which increase bigram entity match by 38.6% and trigram entity match by 41%. Our lexicon, which adds a significant number of new terms, is very useful for matching patients to clinical trials automatically based on eligibility matching. Beyond clinical trial matching, the specialized lexicon developed in this study could serve as a foundation for future healthcare text mining applications.

Expansão automática de léxico para Análise de Sentimentos de Twitter no domínio do Mercado Financeiro Brasileiro

“…Avaliac ¸ão dos l éxicos de sentimento financeiro na classificac ¸ão de tweets no conjunto de dados relacionados ao Mercado Financeiro Brasileiro (em %, melhores valores em negrito). Todas as variac ¸ões do léxico foram comparadas com uma abordagem de préprocessamento dos termos, onde as palavras tanto no dicionário proposto quanto nos tweets a serem classificados foram lematizadas, reduzindo-as ao seu lema raiz [Jung et al 2021]. O melhor resultado foi obtido na proposta S+S/A+PMI (lematizado), com F1-Score de 71,5%, e uma precisão de 71,8%.…”

“…A abordagem lexical é um recurso presente em várias atividades de processamento de linguagem natural, como análise de sentimentos, classificac ¸ão de textos, recuperac ¸ão de opinião e identificac ¸ão de temas, entre outras. Quando elaborados de forma adequada, os léxicos podem fornecer uma boa capacidade de classificac ¸ão, além de poderem ser utilizados como recursos adicionais aos modelos de aprendizagem de máquina [Oliveira et al 2016] Detectar subjetividades em sentenc ¸as e classificálas em uma classe é um desafio, especialmente em domínios específicos, como o mercado de ac ¸ões [Das et al 2022], doenc ¸as [Jung et al 2021], documentos jurídicos [Smywiński-Pohl et al 2019] e outros que exigem corpora especializado.…”

“…Na abordagem lexical, a variac ¸ão dos resultados está ligada à formulac ¸ão do léxico utilizado e seu domínio. Exemplos disso são o estudo de [Jung et al 2021], que cobriu 41% dos termos de vocabulário conhecidos em triagens de câncer de mama. Já [Wang et al 2020] obteve 69,6% de acurácia na análise de sentimentos de comentários de filmes.…”

Mental health at different stages of cancer survival: A natural language processing study of Reddit posts

“…Cancer-related terms were extracted from users' Reddit posts using Cancer.org's glossary of 943 words, which is one of the resources used by Jung and collab. 31 to build a breast cancer lexicon. Similarly, the words related to suicide from Sawhney and collab.…”

Natural Language Processing – Finding the Missing Link for Oncologic Data, 2022

“…history and physical exam) [25,26] , operative reports [3,27], pathology reports [28][29][30][31][32] and imaging reports [33][34][35][36][37][38][39] to analyze outcomes: response [21,[40][41][42], toxicity [43][44][45]and survival [33,46]. Natural Language Processing (NLP) has grown considerably in medicine to be employed for documentation [47,48], outcome prediction [1,27,43,49], phenotyping [50,51], data extraction [17,28,49,[52][53][54][55][56] and analysis, clinical trial eligibility [41,57,58], exploration of literature [51,59,60], evaluating impact on workload and recruitment with active parallel growth in the seconda...…”