Prospective AI-SC Trial

The study will be conducted over an approximate 12-month period during regular weekly UHN Liver Transplant Selection Committee meetings. The primary unit of analysis is the individual liver transplant candidacy decision. The AI-SC output is then presented as an advisory decision-support summary. The committee may maintain or modify its decision and/or rationale. The final decision remains the decision of the human committee.

The study utilizes a pre-post implementation design:

1. Pre-Implementation Phase (Baseline): The committee will conduct evaluations and reach candidacy decisions according to standard clinical workflows without access to the AI tool.
2. Post-Implementation Phase (Intervention): The committee will be presented with the AI-SC recommendations during their routine workflow before finalising the candidacy decision.

Objectives and Evaluated Metrics The primary objective is to evaluate the direct effect of AI-SC recommendations on committee decision-making when deployed prospectively.

To comprehensively assess this impact, the secondary objectives will systematically measure:

• Decision Dynamics: The frequency and specific direction of transplant candidacy decision changes (e.g., from deferral to approval, or vice versa) following the introduction of AI-SC reviews.
• Decision Quality and Rationale: The impact of the tool on the clinical rationale's quality, transparency, completeness, actionability, and perceived fairness.
• Clinician Factors: Multi-disciplinary team members' confidence, acceptance, trust, perceived usefulness, and the overall workflow burden imposed by the tool.
• Predictors of Influence: Patient- and case-specific characteristics that are statistically associated with AI-influenced decision changes.
• Algorithmic Fairness and Equity: The consistency and fairness of the AI tool across diverse patient demographics and clinical profiles, specifically examining sex, race, ethnicity, height, ABO blood group, liver disease etiology, and Area Deprivation Index (ADI).
• Operational Feasibility: System performance metrics including software uptime, error rates, and the tangible impact on weekly meeting duration.
• Clinical Insight and Safety Catchment: Documentation of instances where the AI-SC successfully flags missing clinical data, guideline-relevant issues, potential contraindications, or overlooked clinical opportunities that were not emphasized during the initial multi-disciplinary discussion.

Eligibility Criteria:

Ages Eligible for Study: 18 Years and older (Adult, Older Adult) Sexes Eligible for Study: All Accepts Healthy Volunteers: No

Inclusion Criteria:

• Age 18 years or older.
• Undergoing a liver transplant assessment at UHN.
• Anticipated presentation and discussion at the weekly UHN Liver Transplant Selection Committee meeting.
• Ability to provide informed consent.

Exclusion Criteria:

• Refusal or withdrawal of consent.
• Cases for which required committee presentation information is unavailable to the study team.
• Cases in which urgent clinical circumstances do not allow the informed consent process to occur prior to the scheduled committee review.
• Cases in which the treating clinical team determines that research contact is inappropriate due to severe clinical, emotional, or logistical circumstances.

The AI-SC is an investigational clinical decision-support tool designed to provide a structured advisory recommendation for liver transplant candidacy discussions. It consists of six specialty-specific LLM-based agents representing hepatology, transplant surgery, cardiology, anesthesiology, psychiatry, and social work. The system uses a retrieval-augmented generation framework that references a locked knowledge base of institutional transplant criteria and relevant clinical guidelines.

Time frame: During weekly UHN Liver Transplant Selection Committee meetings for approximately 12 months.

Not reported in this registry record.

Artificial Intelligence-Selection Committee (AI-SC) Trial: A Prospective Study of Human-AI Collaborative Decision-Making in Liver Transplant Selection