Differences in information accessed in a pharmacologic knowledge base using a conversational agent vs traditional...

Area of Science:

• Pharmacology
• Health Informatics
• Artificial Intelligence in Medicine

Background:

• Clinicians utilize pharmacologic knowledge bases for medication information and to prevent adverse drug events.
• Micromedex® integrated Watson Assistant (WA), an AI-powered conversational agent, for natural language medication queries.

Purpose of the Study:

• To evaluate differences in information access frequency between traditional search methods (keyword, navigation) and conversational agent search (WA).
• To compare the proportion of information types accessed via WA versus traditional methods.

Main Methods:

• A comparative analysis of information access proportions was conducted.
• Data was collected during the first six months of 2020, comparing WA usage to traditional search methods within Micromedex®.

Main Results:

• The distribution of accessed information types differed significantly between WA and traditional methods (P < .0001).
• Top accessed information included single drug/dosing, FDA/non-FDA uses, adverse effects, and drug administration for both methods.
• Intravenous compatibility information was accessed more frequently via WA (7.7%) compared to traditional methods (0.6%).
• Dose adjustments were accessed more frequently via traditional methods (4.8%) than WA (1.4%).

Conclusions:

• Significant differences exist in how clinicians access pharmacologic information using conversational agents versus traditional search methods.
• User-centered studies are recommended to understand the observed differences in information access patterns.