Automatic Generation of Conditional Diagnostic Guidelines

Area of Science:

• Medical Informatics
• Natural Language Processing
• Clinical Decision Support

Background:

• Diagnostic workups involve nuanced decisions on procedure appropriateness.
• Reference materials contain extensive, complex information on medical procedures.

Purpose of the Study:

• To develop an automated pipeline for extracting conditional statements regarding medical procedure recommendations.
• To create concise summaries of procedural recommendations based on suspected diagnoses.

Main Methods:

• A three-task pipeline was employed: procedure recommendation identification, conditional statement extraction, and coreference resolution.
• The system analyzes diagnostic text to link procedures with their specific conditions.

Main Results:

• Each task in the pipeline achieved above-baseline performance.
• The system successfully produced concise recommendation summaries.

Conclusions:

• Automated extraction of procedural conditions from medical texts is feasible.
• The developed pipeline can enhance clinical decision-making by providing clear recommendation summaries.