How Do Patients Perceive AI?

Researchers address patient perception and ethical challenges related to the transformative technology


While artificial intelligence (AI) has the potential to transform radiology, many in the specialty may not have stopped to consider how patients feel about use of the technology.

In fact, there is a lack of research overall on how patients view AI in radiology, according to Thomas Kwee, MD, PhD, co-author on a European Radiology study examining the issue.

“A lot of research deals with technical developments and clinical applications of AI in radiology, and, to a lesser extent, ethical issues related to AI,” said Dr. Kwee, Department of Radiology, Medical Image Center, the University Medical Center, Groningen, the Netherlands.

“However, there has been a lack of research on how patients view the use of AI in radiology and how to measure this. Patient participation is relevant, because patients will — directly or indirectly — be impacted by new AI applications, and we as radiologists should consider the patients’ wishes and worries.”

That observation led Dr. Kwee and colleagues to develop and implement a patient questionnaire, starting by conducting semi-structured, qualitative interviews with 20 participants from a previous study. They identified six key domains of patient perspective on AI in radiology: proof of technology, procedural knowledge, competence, efficiency, personal interaction and accountability, and used them as a framework for the questionnaire developed in the study.

Study Identifies Patient Concerns

The questionnaire was completed by 155 patients scheduled for CT, MRI or other conventional radiography from December 2018 to March 2019 at the University Medical Center. Analysis by the researchers yielded five factors that intersect with patients and AI in radiology: distrust and accountability, procedural knowledge, personal interaction, efficiency and being informed.

Dr. Kwee said the average score on the first factor — distrust and accountability — indicates that patients are moderately negative when it comes to their trust in AI in taking over diagnostic interpretation tasks of the radiologist, both with regard to accuracy, communication and confidentiality. (See below.)

Patients Rank AI

Patients in the European Radiology survey named areas of AI that impact them most. Statements in the survey that received the highest ranking from patients in each category.

Distrust and Accountability:

“I find it worrisome that a computer does not take feelings into account.”

Procedural Knowledge:

“I find it important to have a good understanding of the results of a scan.”

Personal Interaction:

“I find it important to ask questions when getting the results.”

Being Informed:

“If a computer would give the results, I would not feel emotional support.”

“We found that patients are generally not overly optimistic about AI systems taking over diagnostic interpretations that are currently performed by radiologists,” Dr. Kwee said.

Patients indicated a preference for personal interaction over AI-based communication and said they would like to be informed and engaged about AI procedures that impact them. Patients also indicated that they would feel a lack of emotional support when computers provide their results versus medical staff.

“Results indicate it is important to actively involve patients when developing AI systems for diagnostic, treatment planning or prognostic purposes, and that patient information and education may be valuable when AI systems with proven value enter clinical practice,” said co-author Yfke Ongena, PhD, assistant professor of communication sciences at the University of Groningen.

Researchers hope the questionnaire will be used by others to measure evolving patient viewpoints on AI applications in radiology and provide valuable information on how to continuously adapt radiological AI systems to the needs of patients.

Radiologists’ Role in Educating Patients About AI

Part of the challenge of educating patients, researchers said, is that AI is still a “black box” and most patients don’t understand what it can or cannot do, or how it will change their health care experience.

Dr. Kwee said radiologists need to be proactive in becoming familiar with AI applications and work with engineers, technical developers, ethical and legal experts and other clinicians to better understand the fast-moving technology.  

“When completely autonomous AI systems are used for diagnostic interpretations, radiologists should take the role of educators to inform both clinicians and patients about the strengths and limitations of these systems,” he said.

The Ethics of AI in Radiology

Aside from educating patients and understanding their concerns, researchers said radiologists also have a role to play in developing an ethical framework for the use of AI in the field.

A number of societies — including RSNA, the American College of Radiology and the European Society of Radiology — issued a joint European and North American Multisociety statement that highlights a consensus that ethical use of AI in radiology should promote well-being, minimize harm, and ensure that all benefits are distributed among stakeholders in a just manner. The statement was published in a 2019 Radiology Special Report led by author J. Raymond Geis, MD, clinical assistant professor of radiology at the University of Colorado School of Medicine, Aurora, and including Prof. Elmar Kotter, MD, MS, consultant in the Department of Radiology, University Hospital Freiburg and Associate Professor of Radiology at Freiburg University, and Drs. Kwee and Ongena.

“The time has come to bring AI innovations to clinical practice,” Dr. Geis said. “However, implementing new AI technology is not a straight-forward process. We also have to consider ethical and legal frameworks.”

Because AI-driven technology requires massive amounts of data, the drive toward commercial access to radiology data is becoming overwhelming. “While this will undoubtedly mature into a reliable and robust infrastructure, we currently lack meaningful experience using such technology for patient care at scale,” Dr. Kotter said.

As the joint statement authors write, “Radiology’s goal should be to derive as much value as possible from the ethical use of AI yet resist the lure of extra monetary gain from unethical uses of radiology data and AI.”

The statement also asserts that AI in radiology should be appropriately transparent, dependable, and curtail bias in decision making. Responsibility and accountability should remain with the humans responsible for patient care, not the technology, according to the authors. In addition, technology and ethics must evolve at the same time.

The authors call on the radiology community to begin to develop codes of ethics and practice for AI that promote use that helps patients and the common good and blocks the use of data solely for financial gain.

“Radiologists are learning about ethical AI at the same time we invent and implement it,” the statement reads. “Technological changes in AI, and society’s response to them, are evolving at a speed and scope that are hard to grasp, let alone manage.”

For More Information

Access the Radiology Special Report, “Ethics of Artificial Intelligence in Radiology: Summary of the Joint European and North American Multisociety Statement.”

Access the European Radiology study, “Patients’ Views on The Implementation of Artificial Intelligence in Radiology: Development and Validation of a Standardized Questionnaire,” at