Artificial intelligence driven framework for the structurization of free-text diagnostic reports
[ACCESS RESTRICTED TO THE UNIVERSITY OF MISSOURI AT REQUEST OF AUTHOR.] Free-text sections of diagnostic reports contain a wealth of data on patients, diseases, and complex diagnostic processes. However, free-text data are a poor starting point for computer-based analytics. The majority of natural language processing (NLP) based approaches lack a capacity to accurately extract complex diagnostic entities and their relationships as well as to provide adequate knowledge representation (KR) for down-stream data mining applications. In order to overcome these limitations, a novel informatics framework is introduced for the analysis of free-text diagnostic reports. The framework is based on artificial intelligence (AI) modeling. Here, AI-based modeling integrates natural language processing information extraction techniques (NLP-IE), ontology-based knowledge representation, n-ary relations according to ontological patterns, and information entropy-based data mining approaches. Diagnostic reports are transformed to knowledge graphs (KGs) of relational triples for further analysis using computers. The goal is to facilitate analysis of diagnostic reports using computers. This informatics framework has potential to broadly impact diagnostic medicine and to be extended to other biomedical domains as well.