Generating recipient-centered explanations about drug prescription

Area of Science:

• Medical Informatics
• Natural Language Generation
• Human-Computer Interaction

Background:

• Knowledge-based systems in drug prescription generate explanations for users.
• Indirect users, distinct from prescribers, are key recipients of these explanations.
• Understanding indirect users' information needs and physician attitudes is crucial for effective explanation generation.

Purpose of the Study:

• To describe the generation of written explanations for indirect users of a drug prescription knowledge-based system.
• To design an Explanation Generator that addresses the specific needs of indirect users.
• To adapt explanation content, order, and style based on the intended indirect user.

Main Methods:

• Utilized studies on indirect users' information needs and physicians' explanatory attitudes.
• Integrated text planning techniques with ATN-based (Augmented Transition Network) surface generation.
• Implemented a double modeling component for adaptive explanation delivery.

Main Results:

• Successfully generated written explanations for indirect users in the drug prescription domain.
• Demonstrated the ability to adapt explanation content, order, and style.
• Provided examples of computer-generated texts for analysis.

Conclusions:

• The developed Explanation Generator offers a method for tailoring explanations to indirect users.
• The approach has potential advantages in communicating complex medical information.
• Further analysis is needed to fully understand the advantages and limitations of this method.