A Bayesian network coding scheme for annotating biomedical information presented to genetic counseling clients.

Area of Science:

• Biomedical Informatics
• Computational Linguistics
• Genetics Communication

Background:

• Layperson-oriented clinical genetics documents often lack structured representation of complex genetic concepts.
• Understanding probabilistic and causal relationships is crucial for effective patient communication in genetics.

Purpose of the Study:

• To develop and evaluate a Bayesian network coding scheme for annotating biomedical content in layperson-oriented clinical genetics documents.
• To support knowledge acquisition for natural language generation projects in clinical genetics.

Main Methods:

• Development of a Bayesian network coding scheme.
• Annotation of a corpus of genetic counseling patient letters.
• Evaluation of intercoder reliability for the developed tag set.

Main Results:

• The coding scheme effectively represents probabilistic and causal relationships in clinical genetics.
• High intercoder reliability was achieved for the tag set, indicating scheme's consistency.
• Demonstrated utility in analyzing discourse and linguistic features in patient-oriented genetic texts.

Conclusions:

• The Bayesian network coding scheme provides a robust method for annotating layperson-oriented clinical genetics content.
• The scheme facilitates deeper analysis of medical communication and supports natural language generation.
• Potential applications include improving patient understanding and analyzing dialogue in medical settings.