Comparing Natural Language Processing and Structured Medical Data to Develop a Computable Phenotype for Patients Hospitalized Due to COVID-19: Retrospective Analysis
Abstract BackgroundThroughout the COVID-19 pandemic, many hospitals conducted routine testing of hospitalized patients for SARS-CoV-2 infection upon admission. Some of these patients are admitted for reasons unrelated to COVID-19 and incidentally test positive for the virus. Because COVID-19–relate...
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
JMIR Publications
2023-08-01
|
| Series: | JMIR Medical Informatics |
| Online Access: | https://doi.org/10.2196/46267 |
| _version_ | 1797733610221993984 |
|---|---|
| author | Feier Chang Jay Krishnan Jillian H Hurst Michael E Yarrington Deverick J Anderson Emily C O'Brien Benjamin A Goldstein |
| author_facet | Feier Chang Jay Krishnan Jillian H Hurst Michael E Yarrington Deverick J Anderson Emily C O'Brien Benjamin A Goldstein |
| author_sort | Feier Chang |
| collection | DOAJ |
| description |
Abstract
BackgroundThroughout the COVID-19 pandemic, many hospitals conducted routine testing of hospitalized patients for SARS-CoV-2 infection upon admission. Some of these patients are admitted for reasons unrelated to COVID-19 and incidentally test positive for the virus. Because COVID-19–related hospitalizations have become a critical public health indicator, it is important to identify patients who are hospitalized because of COVID-19 as opposed to those who are admitted for other indications.
ObjectiveWe compared the performance of different computable phenotype definitions for COVID-19 hospitalizations that use different types of data from electronic health records (EHRs), including structured EHR data elements, clinical notes, or a combination of both data types.
MethodsWe conducted a retrospective data analysis, using clinician chart review–based validation at a large academic medical center. We reviewed and analyzed the charts of 586 hospitalized individuals who tested positive for SARS-CoV-2 in January 2022. We used LASSO (least absolute shrinkage and selection operator) regression and random forests to fit classification algorithms that incorporated structured EHR data elements, clinical notes, or a combination of structured data and clinical notes. We used natural language processing to incorporate data from clinical notes. The performance of each model was evaluated based on the area under the receiver operator characteristic curve (AUROC) and an associated decision rule based on sensitivity and positive predictive value. We also identified top words and clinical indicators of COVID-19–specific hospitalization and assessed the impact of different phenotyping strategies on estimated hospital outcome metrics.
ResultsBased on a chart review, 38.2% (224/586) of patients were determined to have been hospitalized for reasons other than COVID-19, despite having tested positive for SARS-CoV-2. A computable phenotype that used clinical notes had significantly better discrimination than one that used structured EHR data elements (AUROC: 0.894 vs 0.841; PP
ConclusionsThese findings highlight the importance of cause-specific phenotyping for COVID-19 hospitalizations. More generally, this work demonstrates the utility of natural language processing approaches for deriving information related to patient hospitalizations in cases where there may be multiple conditions that could serve as the primary indication for hospitalization. |
| first_indexed | 2024-03-12T12:31:39Z |
| format | Article |
| id | doaj.art-5290c2efd0e2416682616744aabbfbb4 |
| institution | Directory Open Access Journal |
| issn | 2291-9694 |
| language | English |
| last_indexed | 2024-03-12T12:31:39Z |
| publishDate | 2023-08-01 |
| publisher | JMIR Publications |
| record_format | Article |
| series | JMIR Medical Informatics |
| spelling | doaj.art-5290c2efd0e2416682616744aabbfbb42023-08-29T10:08:01ZengJMIR PublicationsJMIR Medical Informatics2291-96942023-08-0111e46267e4626710.2196/46267Comparing Natural Language Processing and Structured Medical Data to Develop a Computable Phenotype for Patients Hospitalized Due to COVID-19: Retrospective AnalysisFeier Chang0http://orcid.org/0009-0009-2535-4810Jay Krishnan1http://orcid.org/0009-0008-6791-4671Jillian H Hurst2http://orcid.org/0000-0001-5079-9920Michael E Yarrington3http://orcid.org/0000-0003-3186-1519Deverick J Anderson4http://orcid.org/0000-0001-6882-5496Emily C O'Brien5http://orcid.org/0000-0002-8257-7561Benjamin A Goldstein6http://orcid.org/0000-0001-5261-3632Duke UniversityDuke UniversityDuke UniversityDuke UniversityDuke UniversityDuke UniversityDuke University Abstract BackgroundThroughout the COVID-19 pandemic, many hospitals conducted routine testing of hospitalized patients for SARS-CoV-2 infection upon admission. Some of these patients are admitted for reasons unrelated to COVID-19 and incidentally test positive for the virus. Because COVID-19–related hospitalizations have become a critical public health indicator, it is important to identify patients who are hospitalized because of COVID-19 as opposed to those who are admitted for other indications. ObjectiveWe compared the performance of different computable phenotype definitions for COVID-19 hospitalizations that use different types of data from electronic health records (EHRs), including structured EHR data elements, clinical notes, or a combination of both data types. MethodsWe conducted a retrospective data analysis, using clinician chart review–based validation at a large academic medical center. We reviewed and analyzed the charts of 586 hospitalized individuals who tested positive for SARS-CoV-2 in January 2022. We used LASSO (least absolute shrinkage and selection operator) regression and random forests to fit classification algorithms that incorporated structured EHR data elements, clinical notes, or a combination of structured data and clinical notes. We used natural language processing to incorporate data from clinical notes. The performance of each model was evaluated based on the area under the receiver operator characteristic curve (AUROC) and an associated decision rule based on sensitivity and positive predictive value. We also identified top words and clinical indicators of COVID-19–specific hospitalization and assessed the impact of different phenotyping strategies on estimated hospital outcome metrics. ResultsBased on a chart review, 38.2% (224/586) of patients were determined to have been hospitalized for reasons other than COVID-19, despite having tested positive for SARS-CoV-2. A computable phenotype that used clinical notes had significantly better discrimination than one that used structured EHR data elements (AUROC: 0.894 vs 0.841; PP ConclusionsThese findings highlight the importance of cause-specific phenotyping for COVID-19 hospitalizations. More generally, this work demonstrates the utility of natural language processing approaches for deriving information related to patient hospitalizations in cases where there may be multiple conditions that could serve as the primary indication for hospitalization.https://doi.org/10.2196/46267 |
| spellingShingle | Feier Chang Jay Krishnan Jillian H Hurst Michael E Yarrington Deverick J Anderson Emily C O'Brien Benjamin A Goldstein Comparing Natural Language Processing and Structured Medical Data to Develop a Computable Phenotype for Patients Hospitalized Due to COVID-19: Retrospective Analysis JMIR Medical Informatics |
| title | Comparing Natural Language Processing and Structured Medical Data to Develop a Computable Phenotype for Patients Hospitalized Due to COVID-19: Retrospective Analysis |
| title_full | Comparing Natural Language Processing and Structured Medical Data to Develop a Computable Phenotype for Patients Hospitalized Due to COVID-19: Retrospective Analysis |
| title_fullStr | Comparing Natural Language Processing and Structured Medical Data to Develop a Computable Phenotype for Patients Hospitalized Due to COVID-19: Retrospective Analysis |
| title_full_unstemmed | Comparing Natural Language Processing and Structured Medical Data to Develop a Computable Phenotype for Patients Hospitalized Due to COVID-19: Retrospective Analysis |
| title_short | Comparing Natural Language Processing and Structured Medical Data to Develop a Computable Phenotype for Patients Hospitalized Due to COVID-19: Retrospective Analysis |
| title_sort | comparing natural language processing and structured medical data to develop a computable phenotype for patients hospitalized due to covid 19 retrospective analysis |
| url | https://doi.org/10.2196/46267 |
| work_keys_str_mv | AT feierchang comparingnaturallanguageprocessingandstructuredmedicaldatatodevelopacomputablephenotypeforpatientshospitalizedduetocovid19retrospectiveanalysis AT jaykrishnan comparingnaturallanguageprocessingandstructuredmedicaldatatodevelopacomputablephenotypeforpatientshospitalizedduetocovid19retrospectiveanalysis AT jillianhhurst comparingnaturallanguageprocessingandstructuredmedicaldatatodevelopacomputablephenotypeforpatientshospitalizedduetocovid19retrospectiveanalysis AT michaeleyarrington comparingnaturallanguageprocessingandstructuredmedicaldatatodevelopacomputablephenotypeforpatientshospitalizedduetocovid19retrospectiveanalysis AT deverickjanderson comparingnaturallanguageprocessingandstructuredmedicaldatatodevelopacomputablephenotypeforpatientshospitalizedduetocovid19retrospectiveanalysis AT emilycobrien comparingnaturallanguageprocessingandstructuredmedicaldatatodevelopacomputablephenotypeforpatientshospitalizedduetocovid19retrospectiveanalysis AT benjaminagoldstein comparingnaturallanguageprocessingandstructuredmedicaldatatodevelopacomputablephenotypeforpatientshospitalizedduetocovid19retrospectiveanalysis |