End user experience of a widely used artificial intelligence based sepsis system

Area of Science:

• Clinical Informatics
• Human-Computer Interaction
• Healthcare Systems Engineering

Background:

• Research on the Epic Sepsis System (ESS) has primarily focused on technical accuracy, overlooking the user experience of healthcare professionals.
• Understanding user experiences is critical for designing effective Artificial Intelligence (AI) systems in clinical settings.

Purpose of the Study:

• To explore the socio-technical dynamics influencing the adoption and use of the Epic Sepsis System (ESS).
• To understand user perceptions and experiences with the ESS in a clinical environment.

Main Methods:

• Purposive sampling was used to interview resident doctors and nurses with recent ESS interaction until data saturation.
• Content analysis was performed using Dedoose software, with codes derived from established frameworks and supplemented by inductive coding.

Main Results:

• Interviews with 10 healthcare providers indicated mixed, though generally positive or neutral, perceptions of the ESS.
• Key themes included workflow integration and usability, with findings organized into workflow fit and usability/utility domains.
• The system demonstrated seamless electronic health record integration, but design gaps were also identified.

Conclusions:

• Clinician experiences with the ESS are influenced by socio-technical factors, impacting adoption and effective use.
• While generally positive, perceptions varied based on professional experience and frequency of system interaction.
• Ongoing refinement of the ESS is necessary, balancing technological progress with clinical practicality to improve future clinical AI tools.