Aligning lay and specialized passages in comparable medical corpora

Area of Science:

• Medical Informatics
• Computational Linguistics
• Health Communication

Background:

• Public access to medical information is increasing.
• Specialized medical terminology presents a barrier for non-experts.
• A need exists to translate complex medical language into lay terms.

Purpose of the Study:

• To develop a method for creating a comparable corpus of expert and non-expert French medical documents.
• To identify similar text segments between lay and specialized medical language.
• To evaluate the effectiveness of the developed method for bridging the language gap.

Main Methods:

• Construction of a comparable corpus of French medical texts from expert and lay sources.
• Application of text-matching algorithms to identify similar segments across the corpus.
• Manual evaluation of retrieved text segment pairs for similarity and exploitability.

Main Results:

• The method successfully identified similar text segments between specialized and lay medical French.
• Out of the top 400 retrieved pairs, 59% were found to be genuinely similar.
• A significant portion (37%) of similar pairs were deemed exploitable for further natural language processing tasks.

Conclusions:

• The developed method provides a viable approach for identifying equivalent expressions in specialized and lay medical language.
• This research offers encouraging evidence for advancing the automatic translation of medical information.
• The findings support efforts to improve health literacy and patient understanding of medical content.