Patients show cautious support for AI in mammography screening
A new study reveals that while patient acceptance of artificial intelligence in breast cancer screening is growing, trust varies based on personal medical history and demographics.
Patients show cautious support for AI in mammography screening | Image Credit: © karelnoppe - © karelnoppe - stock.adobe.com.
There is patient support for artificial intelligence (AI) implementation in screening mammography, according to a recent study published in Radiology: Imaging Cancer, a journal of the Radiological Society of North America.1
The Importance of patient perspectives
Adoption and acceptance of AI technology remains minimal despite improvements in diagnostic accuracy. However, this new data highlights cautious support, influenced by factors such as personal medical history and sociodemographic characteristics.1
“Patient perspectives are crucial because successful AI implementation in medical imaging depends on trust and acceptance from those we aim to serve,” said Basak E. Dogan, MD, study author and clinical professor at the University of Texas Southwestern Medical Center.1
The study was conducted to determine patients’ opinions and concerns about AI in screening mammography.2 Participants received a 29-item survey in either English or Spanish when presenting for screening between February 2023 and August 2023.
Survey structure and participant characteristics
Six categories were developed for the survey questions. These included proof of technology, procedural knowledge, competence, efficiency, personal interaction, and accountability. Responses were measured on a 5-point Likert-type scale.2
Demographic data included race, ethnicity, age, education level, native language, and annual income. Clinical data included menopausal status, age at first mammography screening, number of first-degree relatives with breast cancer, history of abnormal mammograms, and past breast cancer diagnoses.2
There were 518 participants included in the final analysis, 7.92% of whom were undergoing their initial screening mammography. Being aged 40 to 69 years was reported in 72.8%, non-Hispanic White in 50.6%, premenopausal in 29%, and having college-level education or greater in 67%.2
A prior breast cancer diagnosis was reported in 5.21% of participants, a first-degree relative with breast cancer in 27.2%, abnormal mammography results in 35.7%, and a breast biopsy in 22.8%. Approximately 43% had a yearly income over $100,000.2
AI knowledge and preferences
In terms of AI knowledge, 44.4% of participants responded they had “a little bit” of knowledge, while only 1 stated they were an “AI expert,” and 32.1% had no AI knowledge. When using AI during screening mammograms, 74.1% of participants expressed that consent was necessary.2
A preference for AI as a second reader was reported by 71% of participants, while 4.44% indicated comfort using AI alone to interpret their mammograms. Variations were reported in turnaround preferences, with 57.1% of participants willing to wait several hours to a few days for a radiologist’s review, while 19.5% wanted immediate AI results.2
Of participants responding to questions about additional reading requests, 88.9% requested a radiologist review before follow-up for an AI-interpreted abnormal screening, vs 51.3% of radiologist recall reviews by AI. A belief that AI had much worse efficacy vs radiologists was reported by 5.21% of participants, worse by 21%, same by 43.4%, better by 13.5%, and much better by 1.16%.2
Beliefs and implications
Beliefs about accountability were also reported, with 57.7% of participants believing everyone should be accountable for missed cancer detections by AI, while 14.9% would hold the AI manufacturer responsible. Moderate concern about data privacy was reported by 29.5%, while very or extreme concern was reported by 35.5%.2
Sociodemographic factors influenced patients’ responses, such as household income influencing whether patients deemed permission necessary before using AI. Additionally, non-Hispanic Black patients were less likely to agree with AI use vs non-Hispanic White patients. Overall, investigators found cautious support for AI in screening mammography.2
“This highlights how personal medical history influences trust in AI and radiologists differently, emphasizing the need for personalized AI integration strategies in mammographic screening,” said Dogan.1
References
- Patients support AI as radiologist backup in screening mammography. Radiological Society of North America. April 18, 2025. Accessed April 21, 2025. https://www.eurekalert.org/news-releases/1080589?
- Ozcan B, Dogan BE, Xi Y, Knippa EE. Patient perception of artificial intelligence use in interpretation of screening mammograms: a survey study. Radiology: Imaging Cancer. doi:10.1148/rycan.240290
