96% of Patients Want Disclosure When AI Reads Medical Scans

Patients demand transparency in AI-assisted radiology

Nearly all patients want to be informed when artificial intelligence tools help interpret their medical imaging studies, according to research published in Radiology by the Radiological Society of North America.

The study surveyed more than 1,000 patients at Beaumont Hospital in Dublin, Ireland, between July 2024 and April 2025. Approximately 96% said patients should be told when AI assists in analyzing their scans. Among those respondents, 53% preferred providing written consent while 34% favored verbal notification.

The findings address a gap in existing literature, which has focused primarily on how physicians and medical students view AI in radiology rather than patient perspectives. Most respondents were between ages 50 and 69 and identified as white Irish.

Why it matters

As healthcare systems rapidly deploy AI tools to assist radiologists with image interpretation, the question of patient notification has remained largely unresolved. This research suggests transparency isn't just a regulatory consideration—it's what patients actively expect. Organizations implementing AI in diagnostic workflows may need to balance patient preferences for disclosure against operational constraints, particularly if written consent becomes standard practice.

Accountability concerns surface

When asked who bears responsibility if AI and a radiologist produce an incorrect diagnosis, 64% of respondents said both the physician and the technology should be held accountable. Dr. Hayley Briody and colleagues at Beaumont Hospital's Department of Radiology interpret this as evidence that patients want clarity about how AI functions in their care.

The researchers noted that current regulations don't mandate disclosure of AI tools used for image interpretation. While patients have a right to be informed, obtaining written consent for every AI-assisted reading could prove time-intensive in clinical practice.

The study authors referenced the Blueprint for an "AI Bill of Rights," which emphasizes a "right to notice and explanation" but doesn't specify whether standardized notification or formal informed consent should be required.

Study limitations and future research

The research team acknowledged several constraints. The survey was available only in English, potentially introducing confounding factors. Selection bias may have affected results since patients were recruited from imaging department waiting areas. The total number of patients invited to participate remains unknown, making it impossible to calculate a response rate.

Despite these limitations, Briody and co-authors hope their work encourages additional research into patient attitudes when AI serves secondary functions, such as prioritizing radiology requests based on urgency. They emphasized the importance of including patient representatives in AI governance discussions.

"Overall, despite this study's limitations, transparency appears important when implementing AI tools in radiology to ensure patient comfort with their use," the authors concluded.

The findings were first reported by Radiology Business based on the research published June 30 in the journal Radiology.