Op-Ed: Most Patients Want to Know When Artificial Intelligence Influences Their Care: (Citing Platt et al., 2023, from JAMA Network Open)

 

Publications, Podcasts, and Articles

Imagine walking into a modern hospital, where the bustling corridors are alive with activity and a sense of urgency swarms. Radiology technologists are in flux with patient care; doctors confer about treatment plans, and a radiologist analyzes a CT of an ER patient's exam. What might not be immediately apparent is that the radiologist—like many other healthcare professionals—is assisted by invisible technology, often powered by cutting-edge artificial intelligence (AI). This technology can rapidly sift through troves of medical images, flagging areas of concern before the radiologist has had time to blink. It sounds like something from the future, but AI has become integral in today's healthcare. From routine imaging analysis to personalized treatment plans, AI tools have permeated numerous clinical specialties, particularly radiology. Nevertheless, as the technology becomes more sophisticated, essential questions about patient autonomy, informed consent, and the ethics of disclosure have surfaced. A study published in JAMA Network Open by Platt et al. (2023) examines these exact considerations. Their research focuses on whether patients want to be informed when AI is used in their care. Contrary to common assumptions that complex science and technology might overwhelm patients, most respondents affirmed their desire for transparency. More than two-thirds reported it was “very true” that they wanted explicit notification about AI’s involvement in their treatments, while only 5% expressed disinterest. Through a narrative exploration of these findings, this op-ed argues that transparency about AI use in clinical settings—particularly in radiology—is crucial for preserving trust, upholding ethical standards, and ensuring patients feel respected and empowered in their healthcare journeys.

In many ways, radiology stands at the front lines of the medical AI revolution. Diagnostic imaging such as MRIs, CT scans, and mammograms often involves massive amounts of complex visual data. By harnessing deep learning algorithms, radiologists can quickly identify subtle markers of disease that may evade even the most well-trained specialists. For instance, AI can detect early-stage lung nodules or suspicious calcifications in breast tissue, accelerating the detection of potentially life-threatening conditions and allowing earlier interventions. Yet, while AI-driven tools can undoubtedly improve diagnostic accuracy and efficiency, they also raise concerns about transparency and patient knowledge. Radiology reports may list the final interpretations, providing limited insight into whether a human clinician, a machine, or some combination of the two made the assessment. When so many adult patients in the United States desire to be notified about AI involvement, the typical radiology workflow—where AI might be used behind the scenes—signals a pressing need for systemic disclosure policies.

Conducted in the summer of 2023, the survey by Platt et al. (2023) aimed to gauge real-world attitudes toward AI. Respondents were not simply asked dry questions; they were first shown an informative video outlining the expanding role of AI in healthcare. The goal was to provide a baseline level of understanding so that people’s responses would come from a place of informed reflection. The results spoke volumes. Approximately 67% of participants stated unequivocally that it was “very true” that they wanted to be informed. What may come as a surprise is that only a fraction—less than 5%—voiced reluctance or indifference about such disclosure. This finding counters the narrative that patients might feel overwhelmed or apathetic about the technicalities of their care. Instead, it reinforces the principle that autonomy and respect for individuals’ right to know are cornerstones of ethical healthcare. A secondary layer to the findings concerned demographic distinctions. Women were more likely than men to wish for notification, and White participants were more likely to express interest in AI disclosures than Black respondents. While the study did not investigate the reasons behind these differences, they could reflect social, cultural, or historical factors related to technology trust and experiences with the healthcare system. Regardless of the causes, these demographic nuances underscore the need for health systems to adopt inclusive, culturally sensitive strategies to inform all patients about the role of AI in their care.

The concept of informed consent transcends procedural checklists; it is about honoring patient autonomy. When patients understand what interventions they are receiving, they can more freely decide whether they are comfortable with those methods. Traditionally, informed consent has involved disclosing information about surgical or medicinal interventions, including potential risks and benefits. However, the introduction of AI complicates matters. These algorithms can be viewed as “black boxes,” with complex layers of computation that are not immediately interpretable even to the clinicians who employ them. When a system is so opaque that even experts cannot fully explain how it concluded, conveying meaningful information to patients can become challenging. Nevertheless, as Platt et al. (2023) argue, the notification is necessary for ethical AI implementation. It affirms a “right to know,” ensuring that no patient is unwittingly part of a process where algorithms influence decisions about diagnoses or treatments without their awareness. In radiology, where AI is likely to interpret the images before a radiologist reviews them, this principle should theoretically be as standard as an X-ray technologist informing a patient about the steps involved in an imaging procedure.

A key question is “How” and “When” health systems should alert patients to AI usage. While the findings from Platt et al. (2023) make a compelling case for disclosure, they do not specify how that disclosure should happen. Some experts suggest that a succinct, standardized disclosure could suffice in scenarios where AI is widely accepted, such as automated image analysis for standard chest X-rays. This might involve a simple statement in the patient’s electronic portal indicating that AI analysis aids in diagnosing certain conditions. In contrast, more elaborate explanations and consent procedures may be warranted when AI significantly impacts the care plan—perhaps by helping a physician decide whether to proceed with a risky surgery. Furthermore, these disclosures must be carried out in ways that are culturally and linguistically accessible. For instance, African American communities with historical reasons for distrusting medical institutions might need a more comprehensive conversation about AI and its uses. Similarly, non-English speakers or individuals with lower digital literacy might need multimedia tools or face-to-face discussions rather than lengthy written materials. Tailoring these messages not only addresses ethical obligations but also helps to foster trust and engagement. Although the United States has yet to adopt comprehensive, unified regulations governing the use and disclosure of AI in healthcare, frameworks from other regions might offer guidance. European law, for example, includes articles within the General Data Protection Regulation (GDPR) that stress transparency and the right to an explanation regarding automated decision-making. While these regulations do not strictly apply to all U.S. clinical settings, they underscore the global trend toward holding organizations accountable for disclosing how advanced technologies are used. In the American context, the Food and Drug Administration (FDA) has taken steps to evaluate AI and machine-learning-based medical devices. However, a legislative gap remains concerning the mandatory disclosure of AI use to patients. Initiatives from professional bodies, such as the American College of Radiology, could help bridge this gap by proactively establishing best practices that hospitals and imaging centers can adopt. Patient consent for AI use has become a hot topic in radiology. It is estimated that up to half of imaging facilities already implement AI into their workflows. As the number of FDA-approved AI algorithms approaches 1,000—around 70% of which revolve around imaging—the impetus to address notification is becoming more urgent. The survey by Platt et al. (2023) underscores the strong public’s readiness to be informed. Indeed, people want the conversation about AI, and they deserve it. By prioritizing transparency, healthcare institutions can nurture a culture of trust and collaboration. Technology may power the analysis, but the ultimate decisions remain in the clinician-patient relationship. Patients who feel respected and informed are more likely to engage positively with their care, adhere to treatment recommendations, and participate in long-term follow-up. Conversely, secrecy or opacity around AI usage can contribute to distrust, especially among populations who might already harbor skepticism toward the healthcare system.

The healthcare sector stands at a crossroads as AI continues to evolve rapidly. On the one hand, the potential for improved diagnostic accuracy, more personalized treatment plans, and quicker clinical workflows is profoundly transformative. On the other hand, the path to realizing this potential hinges on whether clinicians and policymakers can maintain an unwavering commitment to patient-centered ethics. The study by Platt et al. (2023) illustrates the desires of a patient population that, by and large, wants to be included in decisions affecting their health. AI might be sophisticated, but transparency does not have to be. Ethical guidelines in this new era could be as simple and direct as letting people know that an algorithm contributed to their diagnostic assessment, explaining its role in plain language, and ensuring they have the chance to ask questions or seek second opinions. If healthcare organizations heed these findings, patient trust can be preserved and strengthened. By normalizing notification and weaving it seamlessly into patient care, hospitals, clinics, and imaging centers can demonstrate that AI is not a secretive, top-down imposition but a modern, collaborative tool designed to enhance human expertise. AI may power the future of medicine, but ensuring that patients remain at the center of every decision will keep healthcare humane, ethically grounded, and worthy of public trust.

References

Platt, J. E., et al. (2023). Attitudes toward mandatory AI notifications among patients in a large health system. JAMA Network Open, 6(9), e2332834. https://pubmed.ncbi.nlm.nih.gov/39661391/