Cеминар ММИТ: Integrating Tools for Retrospective Analysis of Big Collection of Clinical Narratives Докладчик: Assoc. Prof. Dr. Svetla Boytcheva
Assoc. Prof. Dr. Svetla Boytcheva, Bulgarian Academy of Sciences, Sofia, Bulgaria
Title: Integrating Tools for Retrospective Analysis of Big Collection of Clinical Narratives
Speaker: Assoc. Prof. Dr. Svetla Boytcheva, Linguistic Modeling and Knowledge Processing Department, Institute of Information and Communication Technologies, Bulgarian Academy of Sciences, Sofia, Bulgaria
Abstract: Today more than 80% of the patient-related clinical information is stored as free text in the Electronic Health Record systems. During the last decade several Information Extraction system for analysis of clinical narratives were developed – for diagnosis extraction, drugs and dosage identification, recognition of complaints and related events, risk factors, etc. Despite the achievements in this area these systems are difficult to (re)use because most of them, including the associated linguistic resources, are language specific (mainly for English language) and cannot be easily adapted for other languages. Moreover, they are developed either as academic research projects or as commercial software. Usually their results are evaluated on annotated corpora manually tuned to specific tasks, so that the performance assessment is difficult as well. The presentation discussed the automatic generation of Diabetic Register from very large repository of free text clinical documents (currently 262 million pseudonymised outpatient records submitted to the Bulgarian National Health Insurance Fund in 2010-2016 for more than 5 million citizens yearly). The construction relies on advanced automatic analysis of free text information as well as on Business Analytics technologies for storing, maintaining, searching, querying and analyzing big data. Original frequent pattern mining algorithms enable to discovery of complex relations between some disorders (comorbidities) taking into account context information. The experiments confirm some known comorbidities; in addition novel hypotheses for discovery of stable comorbidities were generated. Effective explication of comorbidities can fill knowledge gaps and assist informed clinical decision making. The claim is that the synergy of modern analytics tools transforms a static archive of clinical patient records to a sophisticated software environment for knowledge discovery and prediction.