{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,17]],"date-time":"2026-06-17T00:54:31Z","timestamp":1781657671049,"version":"3.54.5"},"reference-count":50,"publisher":"Oxford University Press (OUP)","issue":"5","content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2015,9,1]]},"abstract":"<jats:title>Abstract<\/jats:title>\n               <jats:p>Objective Extracting medical knowledge from electronic medical records requires automated approaches to combat scalability limitations and selection biases. However, existing machine learning approaches are often regarded by clinicians as black boxes. Moreover, training data for these automated approaches at often sparsely annotated at best. The authors target unsupervised learning for modeling clinical narrative text, aiming at improving both accuracy and interpretability.<\/jats:p>\n               <jats:p>Methods The authors introduce a novel framework named subgraph augmented non-negative tensor factorization (SANTF). In addition to relying on atomic features (e.g., words in clinical narrative text), SANTF automatically mines higher-order features (e.g., relations of lymphoid cells expressing antigens) from clinical narrative text by converting sentences into a graph representation and identifying important subgraphs. The authors compose a tensor using patients, higher-order features, and atomic features as its respective modes. We then apply non-negative tensor factorization to cluster patients, and simultaneously identify latent groups of higher-order features that link to patient clusters, as in clinical guidelines where a panel of immunophenotypic features and laboratory results are used to specify diagnostic criteria.<\/jats:p>\n               <jats:p>Results and Conclusion SANTF demonstrated over 10% improvement in averaged F-measure on patient clustering compared to widely used non-negative matrix factorization (NMF) and k-means clustering methods. Multiple baselines were established by modeling patient data using patient-by-features matrices with different feature configurations and then performing NMF or k-means to cluster patients. Feature analysis identified latent groups of higher-order features that lead to medical insights. We also found that the latent groups of atomic features help to better correlate the latent groups of higher-order features.<\/jats:p>","DOI":"10.1093\/jamia\/ocv016","type":"journal-article","created":{"date-parts":[[2015,4,11]],"date-time":"2015-04-11T03:28:51Z","timestamp":1428722931000},"page":"1009-1019","source":"Crossref","is-referenced-by-count":38,"title":["Subgraph augmented non-negative tensor factorization (SANTF) for modeling clinical narrative text"],"prefix":"10.1093","volume":"22","author":[{"given":"Yuan","family":"Luo","sequence":"first","affiliation":[{"name":"Computer Science and Artificial Intelligence Lab, Massachusetts Institute of Technology"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yu","family":"Xin","sequence":"additional","affiliation":[{"name":"Computer Science and Artificial Intelligence Lab, Massachusetts Institute of Technology"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Ephraim","family":"Hochberg","sequence":"additional","affiliation":[{"name":"Center for Lymphoma, Massachusetts General Hospital and Department of Medicine, Harvard Medical School"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Rohit","family":"Joshi","sequence":"additional","affiliation":[{"name":"Computer Science and Artificial Intelligence Lab, Massachusetts Institute of Technology"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Ozlem","family":"Uzuner","sequence":"additional","affiliation":[{"name":"Department of Information Studies, State University of New York at Albany"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Peter","family":"Szolovits","sequence":"additional","affiliation":[{"name":"Computer Science and Artificial Intelligence Lab, Massachusetts Institute of Technology"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"286","published-online":{"date-parts":[[2015,4,9]]},"reference":[{"key":"2020110612594489100_ocv016-B1"},{"key":"2020110612594489100_ocv016-B2","doi-asserted-by":"crossref","first-page":"158rv11","DOI":"10.1126\/scitranslmed.3003528","article-title":"Computational medicine: translating models to clinical care","volume":"4","author":"Winslow","year":"2012","journal-title":"Sci Transl Med."},{"key":"2020110612594489100_ocv016-B3","doi-asserted-by":"crossref","first-page":"68","DOI":"10.1038\/nm0102-68","article-title":"Diffuse large B-cell lymphoma outcome prediction by gene-expression profiling and supervised machine learning","volume":"8","author":"Shipp","year":"2002","journal-title":"Nat Med."},{"key":"2020110612594489100_ocv016-B4","doi-asserted-by":"crossref","first-page":"301","DOI":"10.1006\/jbin.2001.1029","article-title":"A simple algorithm for identifying negated findings and diseases in discharge summaries","volume":"34","author":"Chapman","year":"2001","journal-title":"J Biomed Informat."},{"key":"2020110612594489100_ocv016-B5","first-page":"349","article-title":"Exploiting semantic relations for literature-based discovery","volume":"2006","author":"Hristovski","year":"2006","journal-title":"AMIA Ann Symp Proc."},{"key":"2020110612594489100_ocv016-B6","doi-asserted-by":"crossref","first-page":"19","DOI":"10.1197\/jamia.M3378","article-title":"MedEx: a medication information extraction system for clinical narratives","volume":"17","author":"Xu","year":"2010","journal-title":"J Am Med Inform Assoc."},{"key":"2020110612594489100_ocv016-B7","first-page":"271","article-title":"Methodology to develop and evaluate a semantic representation for NLP","volume":"2009","author":"Irwin","year":"2009","journal-title":"AMIA Ann Symp Proc."},{"key":"2020110612594489100_ocv016-B8","doi-asserted-by":"crossref","first-page":"e29578","DOI":"10.1371\/journal.pone.0029578","article-title":"Unsupervised analysis of classical biomedical markers: robustness and medical relevance of patient clustering using bioinformatics tools","volume":"7","author":"Gordon","year":"2012","journal-title":"PloS One."},{"key":"2020110612594489100_ocv016-B9","doi-asserted-by":"crossref","first-page":"14863","DOI":"10.1073\/pnas.95.25.14863","article-title":"Cluster analysis and display of genome-wide expression patterns","volume":"95","author":"Eisen","year":"1998","journal-title":"Proc Natl Acad Sci."},{"key":"2020110612594489100_ocv016-B10","first-page":"824","article-title":"Automatic lymphoma classification with sentence subgraph mining from pathology reports","volume":"21","author":"Luo","year":"2014","journal-title":"JAMIA."},{"key":"2020110612594489100_ocv016-B11","doi-asserted-by":"crossref","first-page":"e66341","DOI":"10.1371\/journal.pone.0066341","article-title":"Computational phenotype discovery using unsupervised feature learning over noisy, sparse, and irregular clinical data","volume":"8","author":"Lasko","year":"2013","journal-title":"PloS One."},{"key":"2020110612594489100_ocv016-B12","doi-asserted-by":"crossref","first-page":"361","DOI":"10.1007\/s10618-009-0152-3","article-title":"Temporal pattern discovery in longitudinal electronic patient records","volume":"20","author":"Nor\u00e9n","year":"2010","journal-title":"Data Min Knowl Disc."},{"key":"2020110612594489100_ocv016-B13","doi-asserted-by":"crossref","first-page":"788","DOI":"10.1038\/44565","article-title":"Learning the parts of objects by non-negative matrix factorization","volume":"401","author":"Lee","year":"1999","journal-title":"Nature."},{"key":"2020110612594489100_ocv016-B14","doi-asserted-by":"crossref","first-page":"1108","DOI":"10.1038\/nmeth.2651","article-title":"Network-based stratification of tumor mutations","volume":"10","author":"Hofree","year":"2013","journal-title":"Nat Methods."},{"key":"2020110612594489100_ocv016-B15","doi-asserted-by":"crossref","first-page":"401","DOI":"10.1038\/nature07213","article-title":"Regulatory networks define phenotypic classes of human stem cell lines","volume":"455","author":"M\u00fcller","year":"2008","journal-title":"Nature."},{"key":"2020110612594489100_ocv016-B16","doi-asserted-by":"crossref","first-page":"500","DOI":"10.1038\/nm.2344","article-title":"Subtypes of pancreatic ductal adenocarcinoma and their differing responses to therapy","volume":"17","author":"Collisson","year":"2011","journal-title":"Nat Med."},{"key":"2020110612594489100_ocv016-B17","author":"Wang"},{"key":"2020110612594489100_ocv016-B18","doi-asserted-by":"crossref","first-page":"1495","DOI":"10.1093\/bioinformatics\/btm134","article-title":"Sparse non-negative matrix factorizations via alternating non-negativity-constrained least squares for microarray data analysis","volume":"23","author":"Kim","year":"2007","journal-title":"Bioinformatics."},{"key":"2020110612594489100_ocv016-B19","doi-asserted-by":"crossref","first-page":"4164","DOI":"10.1073\/pnas.0308531101","article-title":"Metagenes and molecular pattern discovery using matrix factorization","volume":"101","author":"Brunet","year":"2004","journal-title":"Proc Natl Acad Sci USA."},{"key":"2020110612594489100_ocv016-B20","doi-asserted-by":"crossref","first-page":"3970","DOI":"10.1093\/bioinformatics\/bti653","article-title":"Improving molecular cancer class discovery through sparse non-negative matrix factorization","volume":"21","author":"Gao","year":"2005","journal-title":"Bioinformatics."},{"key":"2020110612594489100_ocv016-B21","doi-asserted-by":"crossref","first-page":"487","DOI":"10.1038\/ng.2955","article-title":"Association of a germline copy number polymorphism of APOBEC3A and APOBEC3B with burden of putative APOBEC-dependent mutations in breast cancer","volume":"46","author":"Nik-Zainal","year":"2014","journal-title":"Nat Genet."},{"key":"2020110612594489100_ocv016-B22","article-title":"Signatures of mutational processes in human cancer","author":"Alexandrov","year":"2013","journal-title":"Nature."},{"key":"2020110612594489100_ocv016-B23","doi-asserted-by":"crossref","first-page":"279","DOI":"10.1007\/BF02289464","article-title":"Some mathematical notes on three-mode factor analysis","volume":"31","author":"Tucker","year":"1996","journal-title":"Psychometrika."},{"key":"2020110612594489100_ocv016-B24","first-page":"11","article-title":"Incremental tensor analysis: theory and applications","volume":"2","author":"Sun","year":"2008","journal-title":"ACM Trans Knowl Discov Data (TKDD)"},{"key":"2020110612594489100_ocv016-B25","doi-asserted-by":"crossref","first-page":"133","DOI":"10.1007\/BF02296963","article-title":"Uniqueness proof for a family of models sharing features of Tucker\u2019s three-mode factor analysis and PARAFAC\/CANDECOMP","volume":"61","author":"Harshman","year":"1996","journal-title":"Psychometrika."},{"key":"2020110612594489100_ocv016-B26","doi-asserted-by":"crossref","first-page":"18371","DOI":"10.1073\/pnas.0709146104","article-title":"A tensor higher-order singular value decomposition for integrative analysis of DNA microarray data from different studies","volume":"104","author":"Omberg","year":"2007","journal-title":"Proc Natl Acad Sci USA."},{"key":"2020110612594489100_ocv016-B27","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1038\/msb.2009.70","article-title":"Global effects of DNA replication and DNA replication origin activity on eukaryotic gene expression","volume":"5","author":"Omberg","year":"2009","journal-title":"Mol Syst Biol."},{"key":"2020110612594489100_ocv016-B28","doi-asserted-by":"crossref","first-page":"S1","DOI":"10.1186\/1471-2164-12-S2-S1","article-title":"Sublineage structure analysis of Mycobacterium tuberculosis complex strains using multiple-biomarker tensors","volume":"12","author":"Ozcaglar","year":"2011","journal-title":"BMC Genomics."},{"key":"2020110612594489100_ocv016-B29","doi-asserted-by":"crossref","first-page":"63","DOI":"10.1186\/1752-0509-2-63","article-title":"Multiway modeling and analysis in stem cell systems biology","volume":"2","author":"Yener","year":"2008","journal-title":"BMC Syst Biol."},{"key":"2020110612594489100_ocv016-B30","first-page":"791","article-title":"Scenario discovery using nonnegative tensor factorization","volume":"5197","author":"Bader","year":"2008","journal-title":"Progress in Pattern Recognit, Image Anal Appl."},{"key":"2020110612594489100_ocv016-B31","doi-asserted-by":"crossref","first-page":"249","DOI":"10.1007\/s10588-005-5380-5","article-title":"Email surveillance using non-negative matrix factorization","volume":"11","author":"Berry","year":"2005","journal-title":"Comput Math Organ Th."},{"key":"2020110612594489100_ocv016-B32","doi-asserted-by":"crossref","first-page":"373","DOI":"10.1016\/j.ipm.2004.11.005","article-title":"Document clustering using nonnegative matrix factorization","volume":"42","author":"Shahnaz","year":"2006","journal-title":"Inform Process Manag."},{"key":"2020110612594489100_ocv016-B33","doi-asserted-by":"crossref","first-page":"147","DOI":"10.1007\/978-1-84800-046-9_8","article-title":"Discussion tracking in Enron email using PARAFAC","author":"Bader","year":"2008","journal-title":"Survey of Text Mining II."},{"key":"2020110612594489100_ocv016-B34","doi-asserted-by":"crossref","first-page":"455","DOI":"10.1137\/07070111X","article-title":"Tensor decompositions and applications","volume":"51","author":"Kolda","year":"2009","journal-title":"SIAM Rev."},{"key":"2020110612594489100_ocv016-B35","doi-asserted-by":"crossref","first-page":"77","DOI":"10.1016\/j.entcs.2004.12.039","article-title":"The gaston tool for frequent subgraph mining","volume":"127","author":"Nijssen","year":"2005","journal-title":"Electron Notes Theor Comput Sci."},{"key":"2020110612594489100_ocv016-B36","doi-asserted-by":"crossref","first-page":"e60954","DOI":"10.1371\/journal.pone.0060954","article-title":"Approximate subgraph matching-based literature mining for biomedical events and relations","volume":"8","author":"Liu","year":"2013","journal-title":"PloS One."},{"key":"2020110612594489100_ocv016-B37","doi-asserted-by":"crossref","first-page":"302","DOI":"10.1016\/j.knosys.2009.11.010","article-title":"Text classification using graph mining-based feature extraction","volume":"23","author":"Jiang","year":"2010","journal-title":"Knowledge-Based Syst."},{"key":"2020110612594489100_ocv016-B38","article-title":"Learning textual graph patterns to detect causal event relations","author":"Rink","year":"2010","journal-title":"FLAIRS Conference, Daytona Beach, Florida"},{"key":"2020110612594489100_ocv016-B39","author":"Liu"},{"key":"2020110612594489100_ocv016-B40","first-page":"161","article-title":"Frequent subtree mining-an overview","volume":"66","author":"Chi","year":"2005","journal-title":"Fundam Inform."},{"key":"2020110612594489100_ocv016-B41","doi-asserted-by":"crossref","first-page":"75","DOI":"10.1017\/S0269888912000331","article-title":"A survey of frequent subgraph mining algorithms","volume":"28","author":"Jiang","year":"2013","journal-title":"Knowl Eng Rev."},{"key":"2020110612594489100_ocv016-B42","volume-title":"Foundations of Statistical Natural Language Processing","author":"Manning","year":"1999"},{"key":"2020110612594489100_ocv016-B43","first-page":"606","article-title":"On the equivalence of nonnegative matrix factorization and spectral clustering","volume":"5","author":"Ding","year":"2005","journal-title":"SDM."},{"key":"2020110612594489100_ocv016-B44","doi-asserted-by":"crossref","DOI":"10.1017\/CBO9780511809071","volume-title":"Introduction to Information Retrieval","author":"Manning","year":"2008"},{"key":"2020110612594489100_ocv016-B45","doi-asserted-by":"crossref","first-page":"1758","DOI":"10.1137\/120887795","article-title":"A block coordinate descent method for regularized multiconvex optimization with applications to nonnegative tensor factorization and completion","volume":"6","author":"Xu","year":"2013","journal-title":"SIAM J Imaging Sci."},{"key":"2020110612594489100_ocv016-B46","doi-asserted-by":"crossref","first-page":"649","DOI":"10.1016\/j.patcog.2011.05.015","article-title":"Sparse non-negative tensor factorization using columnwise coordinate descent","volume":"45","author":"Liu","year":"2012","journal-title":"Pattern Recogn."},{"key":"2020110612594489100_ocv016-B47","doi-asserted-by":"crossref","first-page":"5228","DOI":"10.1073\/pnas.0307752101","article-title":"Finding scientific topics","volume":"101","author":"Griffiths","year":"2004","journal-title":"Proc Natl Acad Sci USA."},{"key":"2020110612594489100_ocv016-B48","first-page":"1185","article-title":"The Indian buffet process: an introduction and review","volume":"12","author":"Griffiths","year":"2011","journal-title":"J Mach Learn Res."},{"key":"2020110612594489100_ocv016-B49","doi-asserted-by":"crossref","first-page":"48","DOI":"10.1126\/scitranslmed.3001304","article-title":"Integration of early physiological responses predicts later illness severity in preterm infants","volume":"2","author":"Saria","year":"2010","journal-title":"Sci Transl. Med."},{"key":"2020110612594489100_ocv016-B50","first-page":"1276","article-title":"Prognostic physiology: modeling patient severity in intensive care units using radial domain folding","volume":"2012","author":"Joshi","year":"2012","journal-title":"AMIA Annu Symp Proc."}],"container-title":["Journal of the American Medical Informatics Association"],"original-title":[],"language":"en","link":[{"URL":"http:\/\/academic.oup.com\/jamia\/article-pdf\/22\/5\/1009\/34146351\/ocv016.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"http:\/\/academic.oup.com\/jamia\/article-pdf\/22\/5\/1009\/34146351\/ocv016.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2020,11,6]],"date-time":"2020-11-06T18:39:52Z","timestamp":1604687992000},"score":1,"resource":{"primary":{"URL":"https:\/\/academic.oup.com\/jamia\/article\/22\/5\/1009\/929564"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2015,4,9]]},"references-count":50,"journal-issue":{"issue":"5","published-online":{"date-parts":[[2015,4,9]]},"published-print":{"date-parts":[[2015,9,1]]}},"URL":"https:\/\/doi.org\/10.1093\/jamia\/ocv016","relation":{},"ISSN":["1527-974X","1067-5027"],"issn-type":[{"value":"1527-974X","type":"electronic"},{"value":"1067-5027","type":"print"}],"subject":[],"published-other":{"date-parts":[[2015,9]]},"published":{"date-parts":[[2015,4,9]]}}}