The results of a recent survey showed that patients largely support the use of artificial intelligence (AI) to aid radiologists in reading mammograms. The findings, which were published in Breast Cancer Research and Treatment, also indicated acceptance varied in association with factors such as age, race, and education.
“This is the first study to measure patient perspectives on AI in mammography in different hospital settings,” said corresponding author Basak Dogan, MD, a Eugene P. Frenkel, MD Scholar in Clinical Medicine; Professor of Radiology, Director of Breast Imaging Research; and Member of the Harold C. Simmons Comprehensive Cancer Center at UT Southwestern. “It reveals how demographic and socioeconomic factors shape acceptance, trust, and concerns about AI integration in breast cancer screening.”
Methods and Participant Demographics
A 29-item survey regarding patient perceptions of AI use in interpretation of mammograms was conducted at UT Southwestern and offered to all patients who visited the Parkland Health public safety-net health system between April and June 2024, and to patients who visited the William P. Clements Jr. University Hospital breast imaging clinic between February and August 2023. The researchers analyzed differences in views by patient factors with odds ratios (ORs).
Responses to the survey included 518 participants who visited the academic hospital and 406 who visited the safety-net clinic. Those who visited the academic hospital tended to be older; mostly non-Hispanic White; had higher incomes and levels of education; and a higher self-reported knowledge of AI (P < .001 for all factors) compared with visitors to the safety-net clinic.
Key Findings
Most survey respondents (71.5%) accepted the use of AI in reading mammograms, though participants from the safety-net health system were less likely to accept than participants from the academic hospital (OR = 0.71; 95% confidence interval [CI] = 0.53–0.96; P = .02). Only 6.6% of all participants supported AI as the sole reader of mammograms.
Almost 60% of respondents noted that they would rather wait longer for a read from a radiologist than rely on an immediate read from AI, which supported participants' value placed on human oversight. Eighty-four percent of respondents wanted a radiologist to review a mammogram that AI had identified had an abnormality, whereas 44% wanted AI to review scans flagged by radiologists.
Participants from the safety-net health system were more likely, however, to request that AI read the mammogram after a radiologist identified an abnormality on the scan (OR = 1.83; 95% CI = 1.35–2.49; P < .001), to rate AI as equal or better than a radiologist in the detection of cancer (OR = 1.54; 95% CI = 1.12–2.15; P = .01), to and have greater worries about data privacy (OR = 1.87; 95% CI = 1.22–2.93; P = .01) than participants from the academic hospital.
A total of 73.8% of survey respondents noted that they wanted to be informed of or provide consent before AI was used to read their mammograms. At least 80% of participants reported worrying about issues such as privacy, bias, accuracy, transparency, and the doctor-patient relationship.
“As AI is increasingly used in breast imaging interpretation, attention should be paid to educate patients about the role of AI, obtain consent for its use, and provide safeguards to protect data privacy,” said study leader Emily Knippa, MD, Associate Professor of Radiology and a member of the Breast Imaging Division at UT Southwestern.
Factors associated with higher acceptance of AI use included higher education (OR = 1.99; 95% CI = 1.33–2.99; P < .001) and self-reported knowledge of AI (OR = 1.98; 95% CI = 1.38–2.83; P < .001). On the other hand, non-Hispanic Black race was associated with lower acceptance of AI use (OR = 0.40; 95% CI = 0.25–0.65; P < .001).
DISCLOSURE: The study was funded by Dr. Dogan's Eugene P. Frenkel, M.D., Scholar in Clinical Medicine Award from the Simmons Cancer Center. For full disclosures of the study authors, visit link.springer.com.
