Patient Record Linkage for Data Quality Assessment Based on Time Series Matching

Area of Science:

• Health Informatics
• Data Science
• Biomedical Engineering

Background:

• Healthcare generates vast datasets, but data protection regulations restrict secondary use.
• Record linkage is crucial for integrating fragmented datasets lacking universal identifiers.

Purpose of the Study:

• To link two ergometric performance test datasets for data quality assessment.
• To establish reference values for free-text annotations.

Main Methods:

• An iterative, distance-based time series record linkage algorithm was developed in Matlab.
• The algorithm identified corresponding patient and ergometry records across datasets.

Main Results:

• The record linkage algorithm achieved a 74.5% patient record matching rate.
• The highest rate of appropriate free-text annotations reached 87.9%.

Conclusions:

• The developed algorithm successfully linked a significant portion of patient records.
• While lacking a gold standard, results indicate high data quality in free-text annotations.