Automatização do processo de mapeamento de laudos médicos para uma representação estruturada

Abstract

In the health area, large amounts of data from different sources, such as clinical examinations and procedures, are stored in hospitals and medical clinics databases. Characteristics observed in these examinations and procedures are described in textual medical reports. These findings are indispensable for professionals of any medical specialty, as they are used to maintain the clinical history of the patients. The data contained in these reports can be analyzed, in a more complete manner, through the use of computational resources, such as the Data Mining process supported by computational intelligence techniques. However, for the application of such process, it is essential that the data is represented in a structured format, as Computational Data Bases. In this sense, the Laboratory of Bioinformatics from the West Paraná State University in partnership with the Coloproctology Service of the Faculty of Medical Sciences from the State University of Campinas, developed the Mapping Medical Digital Findings Process (PMLM) using ontologies with the aim of providing support for the transformation of unstructured textual medical reports into a structured representation. Nevertheless, the techniques employed in PMLM must be performed manually and separately by means of computer instructions, hampering its use by professionals who are not from the computational area. The objective of this work is to automate and optimize PMLM using ontologies by integrating its preprocessing methods in a computational tool. For this purpose, a Collaborative Computer System (SCC) was built using the prototyping development model, which is applied in five stages: communication, fast plan, modeling, prototype construction, and evaluation and feedback. The developed computational system was evaluated in conjunction with domain experts, confirming that SCC meets the required specifications and presents a great value for the extraction and the study of patterns that may be discovered in medical findings. SCC was also evaluated experimentally, using a set of 100 artificial medical reports that simulate the Upper Digestive Endoscopy (EDA), showing good results. Thus, the automated PMLM using ontologies, implemented in the SCC, enables the performance of more complete and detailed studies, contributing to the generation of new knowledge.