Human-centred AI for emergency cardiac care: Evaluating RAPIDx AI with PROLIFERATE_AI.

Area of Science:

• Medical Informatics
• Artificial Intelligence in Healthcare
• Emergency Medicine

Background:

• Chest pain diagnosis in emergency care presents challenges due to overlapping symptoms, leading to diagnostic uncertainty.
• Artificial Intelligence (AI) tools, such as RAPIDx AI, offer potential to improve diagnostic accuracy by integrating clinical and biochemical data.
• Successful AI adoption requires addressing usability, explainability, and seamless workflow integration within emergency departments.

Purpose of the Study:

• To evaluate the integration of RAPIDx AI into clinical workflows within emergency departments.
• To identify and address usability barriers hindering the adoption of RAPIDx AI.
• To optimize the implementation and use of RAPIDx AI in emergency settings.

Main Methods:

• The PROLIFERATE_AI framework was utilized across 12 emergency departments (EDs) from July 2022 to January 2024.
• Expert Knowledge Elicitation (EKE) informed a survey applied to 24 ED clinicians assessing RAPIDx AI usability and adoption.
• Bayesian inference and Monte Carlo simulations quantified user comprehension, engagement, usage, and preference, while thematic analysis identified barriers and optimization needs.

Main Results:

• Registrars demonstrated the highest comprehension and preference for RAPIDx AI, while residents/interns showed the lowest scores.
• Registered nurses reported strong emotional engagement with the AI tool.
• Novice users encountered usability and workflow integration issues, contrasting with experienced clinicians' suggestions for automation and streamlined processes. RAPIDx AI showed 'Good Impact' with trained users but requires refinement for novices.

Conclusions:

• RAPIDx AI enhances diagnostic accuracy and efficiency for experienced healthcare professionals.
• Usability challenges for novice users necessitate targeted training programs and interface modifications for effective AI integration.
• The PROLIFERATE_AI framework provides a robust method for evaluating and scaling AI solutions in complex healthcare systems.