Patient Feedback Informs New Framework for Clinical Artificial Intelligence Transparency

A new qualitative study across nine hospital sites has established a proportionate transparency framework for implementing artificial intelligence in clinical care, finding that patients view transparency as an essential ethical foundation for autonomy while expecting clinicians to remain the final arbiters of care decisions. Investigating community perspectives through Patient and Family Advisory Councils within a large healthcare network, researchers Guy Stein, Abinav R. Udaiyar, David Bouslov, Howard M. Ross, and Charles E. Binkley analyzed how healthcare systems can translate patient expectations into operational policies. The study addresses a critical operational gap that has emerged as clinical artificial intelligence integration expands rapidly across medicine, ranging from ambient clinical scribes to diagnostic support tools.

Although adoption of artificial intelligence tools by physicians has recently doubled, the study authors observed that empirical research exploring patient expectations regarding agency and transparency remains limited. Furthermore, existing literature has largely failed to provide actionable guidance for health systems seeking to incorporate patient viewpoints into real-world clinical governance. To address these deficiencies, the investigators conducted twenty semi-structured interviews alongside intake surveys with members of Patient and Family Advisory Councils. By sourcing feedback directly from these council members across multiple hospital sites, the research team sought to explore community perspectives on healthcare artificial intelligence and propose a structured model that engenders trust while prioritizing patient-centered care.

To analyze the qualitative interview transcripts, the research team employed a hybrid inductive and deductive thematic approach that combined computational tools with human oversight. Initial coding and theme generation were conducted using Google’s Gemini 2.5 Pro, followed by rigorous human validation to reach an interpretative consensus among the researchers. This analytical process identified four nested themes governing patient perspectives: ethical prerequisites for artificial intelligence integration, systemic prerequisites for trust, preserving clinician oversight, and strategies for meaningful communication.

Within these themes, participants characterized transparency as an indispensable ethical floor that is strictly necessary to preserve patient autonomy. However, the qualitative findings demonstrated that transparency alone is insufficient to secure confidence, as council members expressed that trust must be actively earned through demonstrated evidence of systemic reliability and High Reliability Organization principles. Participants underscored that they expect clinicians to serve as a vital bridge between technology and the patient, firmly retaining authority as the final arbiter of all care decisions.

Drawing directly from these qualitative insights, the researchers formulated a proportionate communication framework that scales transparency according to clinical risk and the necessary degree of patient agency. The resulting operational model stratifies healthcare artificial intelligence engagement into three distinct tiers: disclosure, assent, and consent. By embedding ethical and systemic prerequisites within concrete, actionable steps for proactive communication and robust clinician oversight, the authors conclude that health systems can deploy artificial intelligence technologies in a trust-engendering manner that protects and preserves the sanctity of the patient-provider relationship.

Subject of Research: Patient expectations and transparency frameworks for clinical artificial intelligence

Article Title: Patient expectations for artificial intelligence transparency and agency: A qualitative study and proportionate framework for clinical artificial intelligence implementation

Article References: Stein, G., Udaiyar, A. R., Bouslov, D., Ross, H. M., & Binkley, C. E. (2026). Patient expectations for artificial intelligence transparency and agency: A qualitative study and proportionate framework for clinical artificial intelligence implementation. PLOS Digital Health, 5(11), e0001816. https://doi.org/10.1371/journal.pdig.0001816

Image Credits: AI Generated

DOI: 10.1371/journal.pdig.0001816

Keywords: artificial intelligence, clinical AI, patient autonomy, healthcare transparency, medical ethics, clinical oversight, qualitative research, patient-centered care, Patient, expectations, artificial, intelligence