Avoiding literature overload in the medical domain

Area of Science:

• Medical Informatics
• Information Retrieval

Background:

• Medical literature is vast, hindering patient-related information retrieval for physicians.
• Existing computer systems require time and skills physicians often lack.
• Automated methods can formulate physician information needs but may yield too many results.

Purpose of the Study:

• To assess how to provide physicians with relevant patient literature without extra time or skills.
• To reduce the high number of automatically formulated information needs to prevent information overload.
• To evaluate the effectiveness of different knowledge types in reducing information needs.

Main Methods:

• Developed an automated method to formulate physician information needs.
• Presented four types of knowledge for reducing information needs: temporal, specialization, domain, and user model.
• Conducted an experimental analysis to determine the cause of numerous information needs.

Main Results:

• The number of automatically formulated information needs per physician was high.
• An experimental analysis identified the causes of information need multiplicity.
• Knowledge types were ordered by appropriateness: temporal aspects, physician's specialism, domain knowledge, and user model.

Conclusions:

• Temporal aspects and physician specialization are most effective for reducing information needs.
• Domain knowledge and user models offer additional, though less impactful, reduction strategies.
• Further research is needed to precisely assess the performance of each knowledge type in this domain.