Consumer Understanding of Skin Concerns With an AI-Powered Informational Tool

Area of Science:

• Dermatology
• Artificial Intelligence
• Consumer Health Informatics

Background:

• The impact of artificial intelligence (AI)-powered informational tools on consumer health outcomes remains unclear.
• These tools may influence patient care-seeking behaviors and healthcare system utilization.

Purpose of the Study:

• To evaluate how an AI-powered dermatology application affects consumers' understanding of skin conditions.
• To assess the association between AI tool usage and consumer interpretation of retrospective dermatological cases.

Main Methods:

• A randomized survey study with three arms: control (web search), AI (prototype AI application), and Wizard of Oz (dermatologist ground truth).
• Participants interpreted deidentified skin condition cases, reporting condition identification and appropriate next steps.
• Data collected from US-based participants who had previously sought skin health information.

Main Results:

• Participants using AI and Wizard of Oz tools were more willing to name conditions compared to the control group (62.26% and 61.76% vs. 41.21%).
• AI arm showed increased diagnostic accuracy (22.79%) versus control (7.86%), with Wizard of Oz also showing higher accuracy (36.20%).
• Next-step accuracy improved significantly in the Wizard of Oz group (62.95%) compared to control (60.10%).

Conclusions:

• AI applications enhance consumer accuracy and confidence in understanding skin concerns.
• The degree of accuracy improvement is linked to the AI's prediction accuracy.
• Further design enhancements are necessary to optimize consumer understanding of skin conditions using AI tools.