Artificial intelligence-based clinical decision support for liver transplant evaluation and considerations about...

Area of Science:

• Medical Informatics
• Health Equity
• Transplantation Research

Background:

• Artificial intelligence (AI) and large-scale data use in transplantation decision-making is emerging.
• Transplant candidate assessment is subjective, presenting an opportunity for AI-based clinical decision support (CDS).
• Ensuring fairness and health equity is critical for AI-CDS in transplant applications.

Purpose of the Study:

• To explore healthcare providers' perceptions of AI-CDS for liver transplant listing decisions.
• To identify key considerations for designing fair and effective AI-CDS in transplantation.
• To incorporate human-centered design principles into AI development for transplant care.

Main Methods:

• A multicenter qualitative study was conducted.
• Semistructured interviews were performed with 53 multidisciplinary liver transplant providers.
• Inductive coding and constant comparison analysis were used to analyze interview data.

Main Results:

• Six themes emerged regarding fair AI-CDS design: transparency, interpretability, bias mitigation, AI as a team member, resource needs identification, and patient role inclusion.
• Providers emphasized the need for AI-CDS to be understandable and integrated into the clinical workflow.
• Key design elements focus on ethical considerations and collaborative integration into patient care.

Conclusions:

• Liver transplant providers are cautiously optimistic about AI-CDS potential for improving outcomes and equity.
• Findings offer guidance for designing AI-CDS that prioritizes health equity in liver transplantation.
• Human-centered design is crucial for developing trustworthy and effective AI tools in complex medical fields.