User-Centered Delivery of AI-Powered Health Care Technologies in Clinical Settings: Mixed Methods Case Study

Area of Science:

• Health Informatics
• Human-Computer Interaction
• Artificial Intelligence in Medicine

Background:

• Electronic health record (EHR) technology presents challenges for providers due to time-consuming and effortful tasks.
• Artificial intelligence (AI) enhancements are increasingly deployed in EHRs to address these issues.
• AI implementations in EHRs often lack crucial user-centered evaluation.

Purpose of the Study:

• To evaluate the implementation of an AI-powered search and clinical discovery tool within an EHR system.
• To assess the effectiveness of a user-centered approach in deploying AI tools in clinical settings.

Main Methods:

• A 5-month mixed-methods study was conducted.
• Data collection involved interviews, observations, and surveys.
• User-centered research methods were prioritized throughout the study.

Main Results:

• AI-based EHR features achieved high adoption rates (93% of users within 3 months).
• Significant improvements were observed in user satisfaction and perceived time savings (P<.001).
• The AI tool was successfully integrated into clinical workflows, enhancing user experience.

Conclusions:

• A user-centered approach is feasible and effective for deploying clinical AI tools.
• Close collaboration with users, rapid feedback incorporation, and tailored training drove adoption and positive experiences.
• This study provides a framework for human-centered research methods in AI tool deployment within healthcare.