Derivation and validation of a machine learning record linkage algorithm between emergency medical services and the emergency department
| Autor: | Colby S. Redfield, Yoni Halpern, David Sontag, Edward Ullman, Steven Horng, David W. Schoenfeld, Larry A. Nathanson, A. Tlimat |
|---|---|
| Rok vydání: | 2019 |
| Předmět: |
Male
Emergency Medical Services Quality management Computer science Health Informatics Research and Applications Machine learning computer.software_genre Health informatics Emergency medical services Humans Retrospective Studies business.industry Emergency department Gold standard (test) Social Security number Inter-rater reliability Logistic Models Female Medical Record Linkage Supervised Machine Learning Artificial intelligence Emergency Service Hospital business computer Algorithm Algorithms Record linkage |
| Zdroj: | J Am Med Inform Assoc |
| ISSN: | 1527-974X |
| DOI: | 10.1093/jamia/ocz176 |
| Popis: | ObjectiveLinking emergency medical services (EMS) electronic patient care reports (ePCRs) to emergency department (ED) records can provide clinicians access to vital information that can alter management. It can also create rich databases for research and quality improvement. Unfortunately, previous attempts at ePCR and ED record linkage have had limited success. In this study, we use supervised machine learning to derive and validate an automated record linkage algorithm between EMS ePCRs and ED records.Materials and MethodsAll consecutive ePCRs from a single EMS provider between June 2013 and June 2015 were included. A primary reviewer matched ePCRs to a list of ED patients to create a gold standard. Age, gender, last name, first name, social security number, and date of birth were extracted. Data were randomly split into 80% training and 20% test datasets. We derived missing indicators, identical indicators, edit distances, and percent differences. A multivariate logistic regression model was trained using 5-fold cross-validation, using label k-fold, L2 regularization, and class reweighting.ResultsA total of 14 032 ePCRs were included in the study. Interrater reliability between the primary and secondary reviewer had a kappa of 0.9. The algorithm had a sensitivity of 99.4%, a positive predictive value of 99.9%, and an area under the receiver-operating characteristic curve of 0.99 in both the training and test datasets. Date-of-birth match had the highest odds ratio of 16.9, followed by last name match (10.6). Social security number match had an odds ratio of 3.8.ConclusionsWe were able to successfully derive and validate a record linkage algorithm from a single EMS ePCR provider to our hospital EMR. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|