Using machine learning to link electronic health records in cancer registries: On the tradeoff between linkage quality and manual effort

Publikationstyp:
Zeitschriftenaufsatz
Metadaten:
Autoren
Autoren-URL
https://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=fis-test-1&SrcAuth=WosAPI&KeyUT=WOS:001204450900001&DestLinkType=FullRecord&DestApp=WOS_CPL
DOI
10.1016/j.ijmedinf.2024.105387
eISSN
1872-8243
Externe Identifier
  • Clarivate Analytics Document Solution ID: OA1Q6
  • PubMed Identifier: 38428200
ISSN
1386-5056
Zeitschrift
INTERNATIONAL JOURNAL OF MEDICAL INFORMATICS
Schlüsselwörter
  • Record linkage
  • Data matching
  • Cancer registry
  • Electronic health records
  • Machine learning
  • Data quality
Artikelnummer
ARTN 105387
Datum der Veröffentlichung
2024
Status
Published
Titel
Using machine learning to link electronic health records in cancer registries: On the tradeoff between linkage quality and manual effort
Sub types
  • Article
Ausgabe der Zeitschrift
185

Datenquelle: Web of Science (Lite)

Andere Metadatenquellen:
Autoren
DOI
10.1016/j.ijmedinf.2024.105387
ISSN
1386-5056
Zeitschrift
International Journal of Medical Informatics
Sprache
en
Artikelnummer
105387
Paginierung
105387 - 105387
Datum der Veröffentlichung
2024
Status
Published
Herausgeber
Elsevier BV
Herausgeber URL
http://dx.doi.org/10.1016/j.ijmedinf.2024.105387
Datum der Datenerfassung
2024
Titel
Using machine learning to link electronic health records in cancer registries: On the tradeoff between linkage quality and manual effort
Ausgabe der Zeitschrift
185

Datenquelle: Crossref

Abstract
<h4>Background</h4>Cancer registries link a large number of electronic health records reported by medical institutions to already registered records of the matching individual and tumor. Records are automatically linked using deterministic and probabilistic approaches; machine learning is rarely used. Records that cannot be matched automatically with sufficient accuracy are typically processed manually. For application, it is important to know how well record linkage approaches match real-world records and how much manual effort is required to achieve the desired linkage quality. We study the task of linking reported records to the matching registered tumor in cancer registries.<h4>Methods</h4>We compare the tradeoff between linkage quality and manual effort of five machine learning methods (logistic regression, random forest, gradient boosting, neural network, and a stacked method) to a deterministic baseline. The record linkage methods are compared in a two-class setting (no-match/ match) and a three-class setting (no-match/ undecided/ match). A cancer registry collected and linked the dataset consisting of categorical variables matching 145,755 reported records with 33,289 registered tumors.<h4>Results</h4>In the two-class setting, the gradient boosting, neural network, and stacked models have higher accuracy and F<sub>1</sub> score (accuracy: 0.968-0.978, F<sub>1</sub> score: 0.983-0.988) than the deterministic baseline (accuracy: 0.964, F<sub>1</sub> score: 0.980) when the same records are manually processed (0.89% of all records). In the three-class setting, these three machine learning methods can automatically process all reported records and still have higher accuracy and F<sub>1</sub> score than the deterministic baseline. The linkage quality of the machine learning methods studied, except for the neural network, increase as the number of manually processed records increases.<h4>Conclusion</h4>Machine learning methods can significantly improve linkage quality and reduce the manual effort required by medical coders to match tumor records in cancer registries compared to a deterministic baseline. Our results help cancer registries estimate how linkage quality increases as more records are manually processed.
Addresses
  • Cancer Registry, Institute for Digital Health Data Rhineland-Palatinate, Große Bleiche 46, Mainz, 55116, Germany; Information Systems and Business Administration, Johannes Gutenberg University, Jakob-Welder-Weg 9, Mainz, 55128, Germany. Electronic address: roechner@uni-mainz.de.
Autoren
DOI
10.1016/j.ijmedinf.2024.105387
eISSN
1872-8243
Externe Identifier
  • PubMed Identifier: 38428200
Open access
false
ISSN
1386-5056
Zeitschrift
International journal of medical informatics
Schlüsselwörter
  • Humans
  • Neoplasms
  • Medical Record Linkage
  • Registries
  • Databases, Factual
  • Electronic Health Records
Sprache
eng
Medium
Print-Electronic
Online publication date
2024
Paginierung
105387
Datum der Veröffentlichung
2024
Status
Published
Datum der Datenerfassung
2024
Titel
Using machine learning to link electronic health records in cancer registries: On the tradeoff between linkage quality and manual effort.
Sub types
  • Journal Article
Ausgabe der Zeitschrift
185

Datenquelle: Europe PubMed Central

Abstract
BACKGROUND: Cancer registries link a large number of electronic health records reported by medical institutions to already registered records of the matching individual and tumor. Records are automatically linked using deterministic and probabilistic approaches; machine learning is rarely used. Records that cannot be matched automatically with sufficient accuracy are typically processed manually. For application, it is important to know how well record linkage approaches match real-world records and how much manual effort is required to achieve the desired linkage quality. We study the task of linking reported records to the matching registered tumor in cancer registries. METHODS: We compare the tradeoff between linkage quality and manual effort of five machine learning methods (logistic regression, random forest, gradient boosting, neural network, and a stacked method) to a deterministic baseline. The record linkage methods are compared in a two-class setting (no-match/ match) and a three-class setting (no-match/ undecided/ match). A cancer registry collected and linked the dataset consisting of categorical variables matching 145,755 reported records with 33,289 registered tumors. RESULTS: In the two-class setting, the gradient boosting, neural network, and stacked models have higher accuracy and F1 score (accuracy: 0.968-0.978, F1 score: 0.983-0.988) than the deterministic baseline (accuracy: 0.964, F1 score: 0.980) when the same records are manually processed (0.89% of all records). In the three-class setting, these three machine learning methods can automatically process all reported records and still have higher accuracy and F1 score than the deterministic baseline. The linkage quality of the machine learning methods studied, except for the neural network, increase as the number of manually processed records increases. CONCLUSION: Machine learning methods can significantly improve linkage quality and reduce the manual effort required by medical coders to match tumor records in cancer registries compared to a deterministic baseline. Our results help cancer registries estimate how linkage quality increases as more records are manually processed.
Date of acceptance
2024
Autoren
Autoren-URL
https://www.ncbi.nlm.nih.gov/pubmed/38428200
DOI
10.1016/j.ijmedinf.2024.105387
eISSN
1872-8243
Zeitschrift
Int J Med Inform
Schlüsselwörter
  • Cancer registry
  • Data matching
  • Data quality
  • Electronic health records
  • Machine learning
  • Record linkage
  • Humans
  • Electronic Health Records
  • Medical Record Linkage
  • Neoplasms
  • Registries
  • Databases, Factual
Sprache
eng
Country
Ireland
Paginierung
105387
PII
S1386-5056(24)00050-9
Datum der Veröffentlichung
2024
Status
Published
Datum, an dem der Datensatz öffentlich gemacht wurde
2024
Titel
Using machine learning to link electronic health records in cancer registries: On the tradeoff between linkage quality and manual effort.
Sub types
  • Journal Article
Ausgabe der Zeitschrift
185

Datenquelle: PubMed

Beziehungen:
Eigentum von

Werkzeuge