Pathology report data extraction from relational database using R, with extraction from reports on melanoma of skin...

Area of Science:

• Pathology Informatics
• Biostatistics
• Computational Pathology

Background:

• Traditional pathology data extraction methods have limitations in success rates.
• Developing efficient data extraction techniques from pathology reports is crucial for research and clinical applications.
• Relational databases are increasingly used for storing pathology information.

Purpose of the Study:

• To describe a technique for directly extracting data from a relational database for pathology reports.
• To demonstrate the application of R scripting for querying and retrieving specific cancer data.
• To evaluate the accuracy and efficiency of data extraction for melanoma staging.

Main Methods:

• Utilized R scripting language with the RODBC package to query a pathology information system database.
• Extracted data from synoptic reports for melanoma of the skin over a 4.5-year period.
• Performed a secondary query to retrieve lymph node staging information from subsequent reports of the same patients.

Main Results:

• Successfully retrieved 426 synoptic reports for melanoma of the skin and extracted key data elements.
• Achieved 82% precise retrieval of lymph node staging information (pN0-pN3) from subsequent reports.
• Demonstrated 100% specificity in retrieving lymph node staging information, with 15% retrieved ambiguously.

Conclusions:

• R scripting with RODBC offers a simple, versatile, and effective method for pathology data extraction from relational databases.
• Data retrieval success is contingent on report formatting and consistent phrasing of data elements.
• This approach is adaptable to various pathology information systems utilizing relational databases.