A computational system based on ontologies to automate the mapping process of medical reports into structured databases. (January 2019)

• Record Type: Journal Article
• Authors: Oliva, Jefferson Tales; Lee, Huei Diana; Spolaôr, Newton; Takaki, Weber Shoity Resende; Coy, Claudio Saddy Rodrigues; Fagundes, João José; Wu, Feng Chung
• Is Part Of: Expert systems with applications. Volume 115(2019)
• Page Start: 37
• Page End: 56
• Publication Date: 2019-01
• Subjects: Text mining -- Natural language processing -- Data mining -- Software engineering -- Ontologies -- Medical reports; Expert systems (Computer science) -- Periodicals; Systèmes experts (Informatique) -- Périodiques; Electronic journals; 006.33
• DOI: 10.1016/j.eswa.2018.08.004
• Languages: English
• ISSNs: 0957-4174

Abstract

Highlights: A computational system was developed according to software engineering prototyping. An original ontology-based Medical Report Mapping Process was automated in this work. The system supports automatic mapping of unstructured texts into structured format. An experimental evaluation was performed in a set of 100 artificial textual reports. Experts found that the system is valuable to extract and analyze textual patterns. Abstract: We have developed, in collaboration with medical and computer experts, the ontology-based Medical Report Mapping Process to support the transformation of unstructured reports into a structured representation. Nevertheless, the techniques employed in this two-phase process must be performed individually and manually by computer instructions, which hinder their use by users not familiar with such language. Thereby, this work proposes a tool to automate and optimize this process by integrating its techniques in a computational system, which was built using a software engineering prototyping approach. This system was experimentally evaluated by applying it to a set of 100 textual reports. The first phase decreased the total number of phrases (853) and words (2520) by 82.25% (48) and 92.70% (184), respectively. In the second phase, 100% of the relevant pieces of information (previously established) present in the reports were transcribed. Also, the second phase was applied, using the same configuration as the first study, in another set with 250 textual reports, resulting in a mapping rate of 99.74%. The unprocessed and unmapped words, regarding both experimental evaluations, were recorded for later inclusion into the ontology. By using this system, efficient and scalable investigations can be performed from medical reports, contributing to generate new knowledge. Also, the system facilitates the definition of these structures due to the feasibility to analyze different sentences in unique phrase sets.