Sheng Yu1, Katherine P Liao2, Stanley Y Shaw3, Vivian S Gainer4, Susanne E Churchill4, Peter Szolovits5, Shawn N Murphy6, Isaac S Kohane7, Tianxi Cai8. 1. Partners HealthCare Personalized Medicine, Boston, MA, USA Brigham and Women's Hospital, Boston, MA, USA Harvard Medical School, Boston, MA, USA syu7@partners.org. 2. Brigham and Women's Hospital, Boston, MA, USA Harvard Medical School, Boston, MA, USA. 3. Massachusetts General Hospital, Boston, MA. 4. Research Computing, Partners HealthCare, Charlestown, MA, USA. 5. Massachusetts Institute of Technology, Cambridge, MA, USA. 6. Massachusetts General Hospital, Boston, MA Research Computing, Partners HealthCare, Charlestown, MA, USA. 7. Harvard Medical School, Boston, MA, USA Boston Children's Hospital, Boston, MA, USA. 8. Harvard T.H. Chan School of Public Health, Boston, MA, USA.
Abstract
OBJECTIVE: Analysis of narrative (text) data from electronic health records (EHRs) can improve population-scale phenotyping for clinical and genetic research. Currently, selection of text features for phenotyping algorithms is slow and laborious, requiring extensive and iterative involvement by domain experts. This paper introduces a method to develop phenotyping algorithms in an unbiased manner by automatically extracting and selecting informative features, which can be comparable to expert-curated ones in classification accuracy. MATERIALS AND METHODS: Comprehensive medical concepts were collected from publicly available knowledge sources in an automated, unbiased fashion. Natural language processing (NLP) revealed the occurrence patterns of these concepts in EHR narrative notes, which enabled selection of informative features for phenotype classification. When combined with additional codified features, a penalized logistic regression model was trained to classify the target phenotype. RESULTS: The authors applied our method to develop algorithms to identify patients with rheumatoid arthritis and coronary artery disease cases among those with rheumatoid arthritis from a large multi-institutional EHR. The area under the receiver operating characteristic curves (AUC) for classifying RA and CAD using models trained with automated features were 0.951 and 0.929, respectively, compared to the AUCs of 0.938 and 0.929 by models trained with expert-curated features. DISCUSSION: Models trained with NLP text features selected through an unbiased, automated procedure achieved comparable or slightly higher accuracy than those trained with expert-curated features. The majority of the selected model features were interpretable. CONCLUSION: The proposed automated feature extraction method, generating highly accurate phenotyping algorithms with improved efficiency, is a significant step toward high-throughput phenotyping.
OBJECTIVE: Analysis of narrative (text) data from electronic health records (EHRs) can improve population-scale phenotyping for clinical and genetic research. Currently, selection of text features for phenotyping algorithms is slow and laborious, requiring extensive and iterative involvement by domain experts. This paper introduces a method to develop phenotyping algorithms in an unbiased manner by automatically extracting and selecting informative features, which can be comparable to expert-curated ones in classification accuracy. MATERIALS AND METHODS: Comprehensive medical concepts were collected from publicly available knowledge sources in an automated, unbiased fashion. Natural language processing (NLP) revealed the occurrence patterns of these concepts in EHR narrative notes, which enabled selection of informative features for phenotype classification. When combined with additional codified features, a penalized logistic regression model was trained to classify the target phenotype. RESULTS: The authors applied our method to develop algorithms to identify patients with rheumatoid arthritis and coronary artery disease cases among those with rheumatoid arthritis from a large multi-institutional EHR. The area under the receiver operating characteristic curves (AUC) for classifying RA and CAD using models trained with automated features were 0.951 and 0.929, respectively, compared to the AUCs of 0.938 and 0.929 by models trained with expert-curated features. DISCUSSION: Models trained with NLP text features selected through an unbiased, automated procedure achieved comparable or slightly higher accuracy than those trained with expert-curated features. The majority of the selected model features were interpretable. CONCLUSION: The proposed automated feature extraction method, generating highly accurate phenotyping algorithms with improved efficiency, is a significant step toward high-throughput phenotyping.
Authors: D H Solomon; N J Goodson; J N Katz; M E Weinblatt; J Avorn; S Setoguchi; C Canning; S Schneeweiss Journal: Ann Rheum Dis Date: 2006-06-22 Impact factor: 19.103
Authors: Katherine M Newton; Peggy L Peissig; Abel Ngo Kho; Suzette J Bielinski; Richard L Berg; Vidhu Choudhary; Melissa Basford; Christopher G Chute; Iftikhar J Kullo; Rongling Li; Jennifer A Pacheco; Luke V Rasmussen; Leslie Spangler; Joshua C Denny Journal: J Am Med Inform Assoc Date: 2013-03-26 Impact factor: 4.497
Authors: Ashwin N Ananthakrishnan; Tianxi Cai; Guergana Savova; Su-Chun Cheng; Pei Chen; Raul Guzman Perez; Vivian S Gainer; Shawn N Murphy; Peter Szolovits; Zongqi Xia; Stanley Shaw; Susanne Churchill; Elizabeth W Karlson; Isaac Kohane; Robert M Plenge; Katherine P Liao Journal: Inflamm Bowel Dis Date: 2013-06 Impact factor: 5.325
Authors: Catherine A McCarty; Rex L Chisholm; Christopher G Chute; Iftikhar J Kullo; Gail P Jarvik; Eric B Larson; Rongling Li; Daniel R Masys; Marylyn D Ritchie; Dan M Roden; Jeffery P Struewing; Wendy A Wolf Journal: BMC Med Genomics Date: 2011-01-26 Impact factor: 3.063
Authors: Joshua C Denny; Marylyn D Ritchie; Melissa A Basford; Jill M Pulley; Lisa Bastarache; Kristin Brown-Gentry; Deede Wang; Dan R Masys; Dan M Roden; Dana C Crawford Journal: Bioinformatics Date: 2010-03-24 Impact factor: 6.937
Authors: Katherine P Liao; Jiehuan Sun; Tianrun A Cai; Nicholas Link; Chuan Hong; Jie Huang; Jennifer E Huffman; Jessica Gronsbell; Yichi Zhang; Yuk-Lam Ho; Victor Castro; Vivian Gainer; Shawn N Murphy; Christopher J O'Donnell; J Michael Gaziano; Kelly Cho; Peter Szolovits; Isaac S Kohane; Sheng Yu; Tianxi Cai Journal: J Am Med Inform Assoc Date: 2019-11-01 Impact factor: 4.497
Authors: Majid Afshar; Dmitriy Dligach; Brihat Sharma; Xiaoyuan Cai; Jason Boyda; Steven Birch; Daniel Valdez; Suzan Zelisko; Cara Joyce; François Modave; Ron Price Journal: J Am Med Inform Assoc Date: 2019-11-01 Impact factor: 4.497
Authors: Benjamin S Glicksberg; Riccardo Miotto; Kipp W Johnson; Khader Shameer; Li Li; Rong Chen; Joel T Dudley Journal: Pac Symp Biocomput Date: 2018