Development of Patient Information Extraction Method by Sequence Labeling using Electronic Medical Records

This research aims to utilize and promote the research and development in the medical field by establishing extraction techniques for patient information such as genetic test results, cancer stage classification, and side effects, which are strongly demanded by pharmaceutical companies. Using two types of methods, "Rule Base (Regular Expression Match)" and "Machine Learning (Sequence Labeling)" with different features as patient information extraction methods, using the Electronic Medical Record (EMR) data of University of Miyazaki Hospital (UMH) was developed. As a result, although it was necessary to evaluate the accuracy of the rule base and machine learning and solve the problem, it was found that the expected patient information could be extracted.