Survey Reveals Patient Support for AI in Mammography Screening

Recent findings from a comprehensive survey indicate that patients are increasingly supportive of integrating artificial intelligence (AI) as a supplementary tool for radiologists in the realm of screening mammography. This study, published in Radiology: Imaging Cancer, highlights the influence of personal medical histories and demographic factors on patients' trust in AI technology.

Despite significant advancements in the diagnostic accuracy of AI systems, widespread acceptance and adoption remain slow due to issues such as data privacy, algorithmic bias, and varying levels of public knowledge regarding AI. A critical yet often overlooked perspective in the discussion of AI's growth within radiology is that of the patient.

Research indicates that patient trust and acceptance are essential for successful AI implementation in medical imaging. A clinical professor of radiology and director of breast imaging research at the University of Texas Southwestern Medical Center emphasized that if patients are skeptical about AI's role in their healthcare, this could adversely affect screening compliance and overall health outcomes.

To explore patient perspectives on AI in mammography, researchers developed a 29-question survey made available to all patients undergoing breast cancer screening at their facility. Conducted over seven months in 2023, the optional survey included closed-ended questions that evaluated participants' knowledge of and attitudes toward AI, alongside demographic and clinical backgrounds, such as previous abnormal mammogram results or family history of breast cancer.

Among the 518 respondents, a significant majority expressed support for utilizing AI in conjunction with radiologist assessments, with 71% preferring AI to serve as a secondary reviewer. However, concerns surrounding the potential loss of personal interaction with radiologists, data privacy, transparency, and algorithmic bias were prevalent, as less than 5% of participants felt comfortable with AI handling their mammogram interpretations independently.

The survey revealed that demographic factors significantly influenced patient attitudes. Individuals with higher education levels or greater self-reported understanding of AI were twice as likely to accept its involvement in their mammography screenings. Notably, Hispanic and non-Hispanic Black respondents demonstrated heightened apprehension regarding AI bias and data privacy, which likely contributed to their lower acceptance levels.

Additionally, family and personal medical histories played a vital role in shaping patient attitudes toward AI. Patients with a relative diagnosed with breast cancer expressed a greater tendency to seek additional reviews, exhibiting a high degree of trust in both AI and radiologist assessments when their mammograms returned normal results. Conversely, those with prior abnormal mammograms were more inclined to pursue further diagnostics if there was a discrepancy between AI and radiologist evaluations, particularly if the AI flagged an abnormality.

This underscores the complex interplay between personal medical experiences and trust in AI versus traditional radiology, highlighting the necessity for tailored AI integration strategies in mammographic screening.

Researchers stress the importance of ongoing engagement with patients to better understand their evolving attitudes toward AI in healthcare, as technology continues to advance. The findings suggest that trust in AI varies significantly among individuals and is influenced by previous medical experiences, educational background, and racial demographics. Incorporating patient insights into AI implementation strategies is crucial to ensuring that these technologies enhance rather than obstruct patient care, ultimately promoting trust and adherence to imaging recommendations.