Data-driven name reduction for record linkage

Centre for Health Record Linkage

International Journal for Population Data Science ◽

10.23889/ijpds.v4i2.1142 ◽

2020 ◽

Vol 4 (2) ◽

Cited By ~ 1

Author(s):

Katie Irvine ◽

Rick Hall ◽

Lee Taylor

Keyword(s):

Record Linkage ◽

Linked Data ◽

Best Practice ◽

Data Linkage ◽

New South ◽

Data Driven ◽

Health Record ◽

Data Governance ◽

South Wales ◽

Data Informed

ContextThe Centre for Health Record Linkage (CHeReL) was established in 2006 as a dedicated health and human services data linkage facility for two Australian jurisdictions, New South Wales and the geographically-nested Australian Capital Territory. The two jurisdictions have their own Governments and separate Health and Human Service systems. Purpose and OperationsThe primary purpose of the CHeReL is to make linked administrative and routinely collected healthdata available to researchers and government within relevant regulatory and governance frameworks.The CHeReL’s data governance and technical operations draw on international best practice andhave been refined by learnings from other data linkage centres. OutcomesOver twelve years of operation, more than 2,320 unique investigators from 140 institutions haveused the CHeReL, producing 615 publications in peer-reviewed literature. A robust pipeline of newdevelopment is expected to further amplify the use of linked data for cutting edge medical researchand support a vision of data-informed policy and data-driven government services.

Download Full-text

Data-adaptive Fellegi-Sunter Model for Probabilistic Record Linkage: Incorporating Missing Data and Field Selection (Preprint)

10.2196/preprints.33775 ◽

2021 ◽

Author(s):

Xiaochun Li ◽

Huiping Xu ◽

Shaun Grannis

Keyword(s):

Public Health ◽

Missing Data ◽

Real World ◽

Record Linkage ◽

Predictive Value ◽

Information Exchange ◽

Missing Values ◽

Missing At Random ◽

Use Cases ◽

Data Driven

UNSTRUCTURED Objectives: We address the real-world challenges of missing data and matching field selection in linking medical records by evaluating the extent to which incorporating the missing-at-random assumption in the Fellegi-Sunter model and using data-driven selected fields improve patient matching accuracy using real-world use cases. Methods: We adapted the Fellegi-Sunter model to accommodate missing data using the missing-at-random assumption and compared the adaptation to the common strategy of treating missing values as disagreement, with matching fields specified by experts or selected by data-driven methods. We used four use cases, each containing a random sample of record pairs with match status ascertained by manual reviews. Use cases included health information exchange (HIE) record deduplication, linkage of public health registry records to HIE, linkage of Social Security Death Master File records to HIE, and deduplicating newborn screening records, which represent real-world clinical and public health scenarios. Matching performance was evaluated using sensitivity, specificity, positive predictive value, negative predictive value, and F-score. Results: Incorporating the missing-at-random assumption in the Fellegi-Sunter model maintained or improved F-scores whether matching fields were expert-specified or selected by data-driven methods. Combining the missing-at-random assumption and data-driven fields optimized F-scores in the four use cases. Conclusions: Missing-at-random is a reasonable assumption in real-world record linkage applications: it maintains or improves F-scores regardless of whether matching fields are expert-specified or data-driven. Data-driven selection of fields coupled with MAR achieves the best overall performance, which can be especially useful in privacy-preserving record linkage.

Download Full-text