Computational Algorithms that Effectively Reduce Report Defects in Surgical Pathology.

Area of Science:

• Medical Informatics
• Computational Pathology
• Health Informatics

Background:

• Pathology reports can contain defects like transcription errors and incorrect non-diagnostic information.
• Examples include incorrect patient demographics, physician details, and tissue block descriptions.
• Computational algorithms have been developed to address these report defects over the last five years.

Purpose of the Study:

• To implement and evaluate computational algorithms for identifying and correcting defects in pathology reports.
• To reduce the incidence of errors in pathology reporting.

Main Methods:

• Algorithms parse gross description texts to identify voice recognition errors using pattern recognition and bigram analysis.
• A separate algorithm compares submitted tissue block information with designated information in the gross description to detect block designation errors.
• Data is retrieved from the pathology information system database.

Main Results:

• Computational algorithms detect voice recognition errors in 8%-10% of cases.
• Block designation errors are identified in approximately 0.5%-1% of all cases.
• The algorithms successfully identified various other report defects.

Conclusions:

• The developed algorithms are effective in reducing pathology report defects.
• These tools can identify voice recognition errors, block designation errors, and other issues like incorrect patient gender or provider information.
• The algorithms contribute to improving the overall quality and accuracy of pathology reports.