A Human-Machine Evaluation of AI in Cardiac Emergencies

Activity: Talk or presentation types › Invited talk

Description

A Human-Machine Evaluation of AI in Cardiac Emergencies
Presenter: Dr. Maria Alejandra Pinero de Plaza

Artificial Intelligence (AI) and Machine Learning (ML) have become pivotal in transforming healthcare, particularly in emergency cardiac care. This presentation, part of a comprehensive evaluation study conducted by Dr. Maria Alejandra Pinero de Plaza and her colleagues, will discuss the preliminary findings of the RAPIDx AI project—a cluster-randomised clinical trial aimed at improving patient outcomes in emergency departments (EDs) through AI integration.

The RAPIDx AI trial involves twelve hospitals across South Australia, where AI tools are employed to assist clinicians in diagnosing and treating patients presenting with chest pain. Dr. Pinero de Plaza’s role focuses on the human-machine interface, specifically evaluating end-user feedback to optimize the AI's implementation.

Summary: "A Human-Machine Evaluation of AI in Cardiac Emergencies" focuses on the integration of Artificial Intelligence (AI) and Machine Learning (ML) in emergency departments (ED) to enhance clinical decision-making and patient outcomes. Key aspects of this research include addressing workflow integration, ensuring seamless data entry, and overcoming interoperability challenges. The RAPIDx AI project is specifically evaluated for its clinical validation, emphasizing the importance of user-centred design and incorporating end-user feedback to optimize its implementation. Ethical considerations and privacy concerns are also critical components in ensuring the responsible deployment of AI in healthcare settings. Expert Knowledge Elicitation (EKE) and Bayesian methods are used to gather comprehensive insights and improve emotional engagement among healthcare professionals.

Key points of the presentation include:

AI Integration in Cardiac Care: Exploring the potential of AI and ML algorithms to enhance decision-making processes and efficiency in cardiovascular health.

Challenges and Barriers: Identifying the primary obstacles to AI adoption, such as workflow integration, data entry, interoperability, and clinical validation. The study also highlights the importance of addressing ethical and privacy concerns associated with AI deployment in healthcare.

PROLIFERATE Framework: Utilizing the PROLIFERATE framework, the study employs a multimethod design combining quantitative and qualitative approaches to gather comprehensive insights. Expert Knowledge Elicitation (EKE) and Bayesian methods are used to quantify expert feedback, providing a robust basis for understanding user acceptance and engagement.

Evaluation Goals: The evaluation aims to ensure that RAPIDx AI meets the practical needs of its users, functions effectively in real-world settings, and continuously improves based on user feedback. This involves assessing comprehension, emotional response, barriers to uptake, and motivation among different user groups, including ED consultants, registrars, residents, and registered nurses.

Findings and Recommendations: Preliminary results indicate moderate comprehension and emotional engagement with RAPIDx AI, with significant variability among different user groups. The study emphasizes the need for comprehensive training programs, continuous feedback loops, and transparent communication about AI functionalities to enhance user trust and engagement. Recommendations for improving AI integration and adoption include hands-on workshops, practical sessions, and simplified data entry processes.

Dr. Pinero de Plaza will conclude with a discussion on the future directions of the RAPIDx AI project, emphasizing the need for large-scale validation, cross-cultural studies, and long-term research to optimize AI integration in clinical settings.
Period11 Jun 2024
Event titleKnowing Exchange Seminar: Caring Futures Institute
Event typeConference
LocationAdelaide, AustraliaShow on map
Degree of RecognitionNational

Keywords

  • Machine Learning (ML)
  • Artificial Intelligence (AI)
  • Cardiac Emergencies
  • Emergency Departments (ED)
  • Workflow Integration
  • Evaluation
  • Implementation Science
  • Expert Knowledge Elicitation (EKE)
  • Human-Machine Interaction

Documents & Links

Related content