Natural language generation of surgical procedures

Area of Science:

• Medical Informatics
• Natural Language Generation
• Computational Linguistics

Background:

• Compositional Medical Concept Representation systems offer detailed information but require natural language translation for end-users.
• The GALEN programme has been developing such representation systems.

Purpose of the Study:

• To report on a tool developed for generating natural language phrases from GALEN conceptual representations.
• To demonstrate the tool's adaptability to different source modeling schemes and destination languages/sublanguages.
• To showcase the application of the tool to model surgical operative procedures within the GALEN-IN-USE project.

Main Methods:

• A multilingual approach to natural language generation with a clear separation of domain and linguistic models.
• Development of specific knowledge structures and operations for bridging conceptual representation and natural language.
• Adaptation of the generation tool to a scheme for modeling surgical procedures, including transformation operations and linguistic knowledge integration.

Main Results:

• The generator successfully adapts to the surgical procedures scheme.
• Demonstrated transformation operations for converting source representations into natural language-translatable forms.
• Generated French phrases for surgical procedures in the urology domain.

Conclusions:

• The developed tool effectively generates natural language from conceptual medical representations.
• The system's modular design allows for adaptation to diverse modeling schemes and languages.
• This facilitates better accessibility and usability of complex medical information for end-users and applications.