Although most patients viewed artificial intelligence as useful and supported its use in health care, they continued to value physician oversight and expressed strong opposition to autonomous decision-making.
Researchers asked 162 patients attending outpatient imaging appointments at a tertiary hospital network in the United Kingdom between September 2023 and January 2024 to complete an 18-item questionnaire designed to assess attitudes toward artificial intelligence (AI) in health care. Exploratory factor analysis was performed to identify underlying themes in patient attitudes, and the researchers also evaluated whether responses differed by demographics.
About 78% of the respondents believed AI would be useful in health care, and 64% believed it should be used. However, 71% of them believed that physicians should remain responsible for decisions involving AI. The researchers identified three principal domains influencing patient attitudes: utility and safety, interaction, and comparability to physicians.
Nearly 70% of the respondents believed AI could reduce health care waiting times, and 77% said they would want to know if AI predicted they might develop a disease in the future. Patients also supported the use of AI to check physicians' clinical judgment rather than replace it.
Despite these favorable views, the respondents consistently preferred physician involvement in care. For instance, 50% of them said they did not trust a computer to make medical decisions, and 70% did not believe physicians could be replaced by artificial intelligence. The respondents also emphasized preserving the human aspects of care: 93% considered it important to be treated as a person rather than "a number," and 80% said they would not be satisfied if AI made decisions without considering their feelings.
In subgroup analyses, female respondents expressed greater reservations about AI compared with male respondents, particularly regarding replacing physician interactions and comparing AI with physicians. The researchers found no statistically significant differences in attitudes according to age or ethnicity.
The researchers acknowledged several limitations. The survey was conducted at a single urban UK hospital network using a convenience sample of outpatient imaging patients, which may limit generalizability to other populations and health care settings. In addition, the questionnaire was intentionally concise to encourage participation and was designed to provide an overview of patient attitudes rather than an in-depth qualitative assessment.
The findings suggested that patients may support the integration of AI into radiology when it complements, rather than replaces, physician expertise and the clinician-patient relationship.
"While there appears to be greater acceptance of the use of AI technology in health care, there is still strong opposition to autonomous AI and any weakening of the clinician-patient relationship," wrote lead study author Rory H. Maclean, BM BCh, of the Department of Cancer Imaging in the School of Biomedical Engineering and Imaging Studies at King's College London, and colleagues.
The study was supported by UK Research & Innovation London Medical Imaging and Artificial Intelligence Centre as well as the Wellcome/Engineering and Physical Sciences Research Council Centre for Medical Engineering at King's College London. Senior study author Vicky Goh, MD , reported institutional research support from Siemens Healthineers. No other conflicts of interest were reported.
Source: British Journal of Radiology
