의무 기록 문서 분류를 위한 자연어 처리에서 최적의 벡터화 방법에 대한 비교 분석
Comparative Analysis of Vectorization Techniques in Electronic Medical Records Classification
2
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
Medical records classification using vectorization techniques plays an important role in natural language processing. The purpose of this study was to investigate proper vectorization techniques for electronic med- ical records classification. Material and methods: 403 electronic medical documents were extracted retrospectively and classified using the cosine similarity calculated by Scikit-learn (Python module for machine learning) in Jupyter Notebook. Vectors for medical documents were produced by three different vectorization techniques (TF-IDF, latent sematic analysis and Word2Vec) and the classification precisions for three vectorization techniques were evaluated. The Kruskal-Wallis test was used to determine if there was a significant difference among three vectorization tech- niques. Results: 403 medical documents were relevant to 41 different diseases and the average number of documents per diagnosis was 9.83 (standard deviation=3.46). The classification precisions for three vectorization techniques were 0.78 (TF-IDF), 0.87 (LSA) and 0.79 (Word2Vec). There was a statistically significant difference among three vec- torization techniques. Conclusions: The results suggest that removing irrelevant information (LSA) is more efficient vectorization technique than modifying weights of vectorization models (TF-IDF, Word2Vec) for medical documents classification.
Abstract
Medical records classification using vectorization techniques plays an important role in natural language processing. The purpose of this study was to investigate proper vectorization techniques for electronic med- ical records classification. Material and methods: 403 electronic medical documents were extracted retrospectively and classified using the cosine similarity calculated by Scikit-learn (Python module for machine learning) in Jupyter Notebook. Vectors for medical documents were produced by three different vectorization techniques (TF-IDF, latent sematic analysis and Word2Vec) and the classification precisions for three vectorization techniques were evaluated. The Kruskal-Wallis test was used to determine if there was a significant difference among three vectorization tech- niques. Results: 403 medical documents were relevant to 41 different diseases and the average number of documents per diagnosis was 9.83 (standard deviation=3.46). The classification precisions for three vectorization techniques were 0.78 (TF-IDF), 0.87 (LSA) and 0.79 (Word2Vec). There was a statistically significant difference among three vec- torization techniques. Conclusions: The results suggest that removing irrelevant information (LSA) is more efficient vectorization technique than modifying weights of vectorization models (TF-IDF, Word2Vec) for medical documents classification.
참고문헌 (26)
- Chicco D, "A machine learning anal-ysis of health records of patients with chronic kidney disease at risk of cardiovascular disease", IEEE Access, 9(3), pp.165132~165144 (2021)
- Blakey JD, "Identifying risk of future asthma attacks using UK medical record data: A respiratory effectiveness group initiative", J Allergy Clin Immunol, 5(4), pp.1015~1024 (2017)
- Tomasallo CD, "Estimating Wiscon-sin asthma prevalence using clinical electronic health records and public health data", Am J Public Health, 104(1), pp.65~73 (2014)
- Spasic I, "Text mining of cancer-related information: Review of current status and future directions", Int J Med Informatics, 83(9), pp.605~623 (2014)
- Jonnalagadda SR, "Text mining of the electronic health record: An information extraction approach for automated identification and subphenotyping of HFpEF patients for clinical trials", J Cardiovasc Transl Res, 10(3), pp.313~321 (2017)
- Rahaman T, "Discovering new trend and connections: Current application of biomedical text mining", Med Ref Services Quarterly, 40(3), pp.329~336 (2021)
- Le Glaz A, "Machine learning and natural language processing in mental health: Systemic review", J Med Internet Res, 23(5), pp.15708~ (2021)
- Shai SS, "Understanding machine learning: from theory to algorithms" (2014), Cambridge University Press
- Peter F, "Machine learning: the art and science of algorithms that make sense of data" (2012), Cambridge University Press
- Mehryar M, "Foundations of machine learning" (2012), MIT press
- Chen MC, "Deep learning to classify radiol-ogy free-text reports", Radiology, 286(3), pp.845~842 (2018)
- 백두현, "텍스트 분류 기반 기계학습의 정신과 진단 예측 적용", 생물정신의학, 27(1), pp.18~26 (2020), 대한생물정신의학회[DOI]
- Andrea C, "Natural language processing and machine learning to enable automatic extraction and classification of patients’ smoking status from electronic medical records", Upsala J Med Sci, 125(4), pp.316~324 (2020)
- Yuli V, "Natural language processing with Python and spaCy:a practical introduction" (2020), No Starch Press
- Hobson L, "Natural language processing in action: understanding, analyzing and generating text with Python" (2019), Manning Publications Co
- Mikolov T, "Efficient estimation of word representations in vector space", Proceedings of Workshop at ICLR, pp.1301~1313 (2013)
- Gastaldi JL, "Why can computers understand natural language? The structuralist image of language behind word embed-dings", Phil Tech, 34(1), pp.149~214 (2021)
- Guillermo JB, "Using latent semantic analysis and the predication algorithm to improve extraction of meanings from a diagnostic corpus", Span J Psychol, 12(2), pp.424~440 (2009)
- Zhou Y, "An introduction to text classification with applica-tions to medical records", 2nd international conference on informa-tional technology and computer application, pp.471~475 (2020)
- Kherwa P, "Latent semantic analysis: an approach to understanding sematic of text", International conference on current trends in computer, electrical, electronics and com-munication, pp.870~874 (2017)
- Almas J, "Affect of data filter on performance of latent semantic analysis based research paper recommender system", 5th International conference on computational intelli-gence and application, pp.50~54 (2020)
- Weng WH, "Medi-cal subdomain classification of clinical notes using a machine learning based natural language processing approach", BMC med inform Decis Mak, 17(1), pp.1~13 (2017)
- Jamaluddin M, "Patient diagnosis classification based on electronic medical record using text mining and support vector machine", International seminar on application for technology of information and communication, pp.243~248 (2021)
- Wang Y, "A clinical text classifica-tion paradigm using weak supervision and deep representa-tion", BMC med inform Decis Mak, 19(1) (2019)
- Park KB, "An empirical study of tokenization strategies for various Korean NLP tasks", Com-puter Science (2020)
- Cho DB, "An empirical study of Korean sentence representation with various tokenization", Electron-ics, 10(7), pp.845~857 (2021)