Aplicação de mineração de texto e processamento de linguagem natural em prontuários eletrônicos de pacientes para extração e transformação de texto em dado estruturado

Abstract

The patient’s medical record is the essential document to ensure comprehensive and continuous care to the patient, providing the history of their health. Over the course of technological evolution, the patient’s therapeutic records have shifted from paper records to the electronic medical record. However, the anamnesis is usually inserted through free text fields, leaving to the health professionals the way information is entered. In this way, traditional Structured Query Language queries are unable to retrieve this data. In order to overcome this problem, we apply Text Mining and Natural Language Processing aiming to extract understandable and standardized data. In this sense, the objective of this work is to propose a tool to recover clinical terms from anamneses and structure them in order to relate to the patterns of pathological diagnosis for later use in complementary studies, thus identifying the most appropriate techniques for the Text Mining process in this context. In this way, a tool was developed to carry out the automatic process of structuring data obtained from anamneses. Thus, after an evaluation between the data collected manually and by the system, the results were subjected to the Kruskal-Wallis statistical test and the hypothesis that there were no significant differences between the samples was accepted.