Translating Medical Dialogues into Pictographs: An Approach Using UMLS

Area of Science:

• Medical Informatics
• Computational Linguistics
• Human-Computer Interaction

Background:

• Medical translation systems require effective ways to represent complex concepts.
• Pictographs offer a potential visual aid for cross-lingual communication in healthcare.
• Existing resources for mapping medical concepts to visual symbols are limited.

Purpose of the Study:

• To explore the feasibility of mapping Unified Medical Language System (UMLS) concepts to pictographs.
• To assess the utility of available pictograph sets for medical domain translation.
• To identify limitations in current pictograph resources for medical concept representation.

Main Methods:

• A systematic attempt was made to map a selection of UMLS concepts to pictographs.
• Two publicly available pictograph datasets were evaluated for concept coverage.
• The effectiveness of word-based lookup strategies was assessed for concept-to-pictograph mapping.

Main Results:

• A significant number of UMLS concepts could not be matched with existing pictographs.
• The coverage of medical concepts within the evaluated pictograph sets was found to be insufficient.
• Word-based retrieval methods proved inadequate for accurately mapping concepts to relevant pictographs.

Conclusions:

• Current pictograph resources are not sufficiently comprehensive for mapping a wide range of UMLS concepts.
• Developing effective visual resources for medical translation requires addressing the gap in concept coverage.
• Further research is needed to improve methods for mapping medical concepts to visual representations.