Structured Processing Of Medical Ultrasound Text Data Based On Semantic Dependency Analysis

In recent years,a series of medical systems such as hospital information systems(HIS),medical imaging systems(PACS),electronic medical records(EHR),laboratory information systems(LIS)and radiological information management systems(RIS)have been producing a lot of data all the time.The data grows infinitely from the initial MB to GB,TB,PB,but the problem of the effective utilization of these medical big data has not been solved.In addition,there are high requirements for real-time and effectiveness in the processing of big data,which cannot be dealt with by traditional analysis methods.As a key medical information carrier,clinical medical text reports provide powerful data support for doctors’ diagnosis and scientific research.However,medical text reports written in natural language are basically unstructured and cannot be directly used for computer analysis and processing.The clinical medical text data has a strong professionalism,which involves many professional medical knowledges,and uses a fixed format in the field of grammar,which makes the extraction of information facing great problems.In information extraction,keyword extraction has a large number of applications in the field of natural language processing.How to quickly and accurately extract keywords from text has become a critical problem to be solved in text processing.There are many existing keyword extraction methods,but there is no keyword extraction method for the medical field.The accuracy and versatility of traditional keyword extraction methods used in the medical field still need to be improved.For this reason,this paper proposes a method for structured expression of clinical medical text data.This method first obtains the professional medical terms in the medical description language through the word segmentation correction method based on the word co-occurrence probability,and then uses the generated professional medical terminology database to provide help for the new round of Chinese word segmentation operations,so that the quality of the word segmentation is significant promoted.Then,the construction of the dependent grammar tree is based on the semantic relationship between the words in a single sentence.Finally,the important indexes and corresponding index values in the medical text are identified and extracted from the syntax tree to obtain structured key-value pair data.The experimental data in this paper use real ultrasound text data.The experimental results show that the word segmentation correction method can greatly improve the quality of word segmentation in Chinese medical texts,with an accuracy rate of 97.4%,and 84.2% accuracy and 87.1% recall in the final structured representation.The structured representation method proposed in this paper can recognize various dependent grammars in medical texts and has good versatility.