“Wicked Problems”: At RSNA, a Radiologist Leader Dissects Process Issues Around AI

Nov. 27, 2023
Elizabeth S. Burnside, M.D., M.P.H., a radiology leader at the University of Wisconsin, in her plenary address on Monday at RSNA23, walked her audience through the process perils around AI

Elizabeth S. Burnside, M.D., M.P.H., senior associate dean in the School of Medicine and Public Health at the University of Wisconsin-Madison, and deputy director of the Institute of Clinical Translational Science for Breast Imaging, at the University of Wisconsin, delivered the first plenary address of the day on Nov. 27 at RSNA23, being held at Chicago’s McCormick Place. Dr. Burnside offered her perspectives as a researcher and clinician, in an address entitled “Leading Through Technology: Valuing Artificial and Human Intelligence,” on Monday morning.

In order to immediately engage her audience, Burnside began by sharing on the screen overhead a little experiment she performed using generative AI. She decided she wanted to try to leverage generative AI to create images of Wilhelm Röntgen (1845-1923), the German mechanical engineer and physicist who discovered X-rays; Marie Curie (1867-1964), the Polish-French physicist and chemist who won the Nobel Prize twice, for her pioneering work in radioactivity; and Sir Godfrey Hounsfield (1919-2004), a British electrical engineer who won the Nobel Prize, at Wrigley Field, the baseball stadium in Chicago. Using generative AI, Burnside attempted to create a picture of the three scientists standing together at Wrigley Field, only to end up with Hounsfield having three legs; next, she created an image of the three standing in front of The Bean, the famous Chicago sculpture. It was witty and amusing, and she was able to quickly convey a tiny sense of the challenge involved in leveraging generative AI.

More seriously, Burnside next turned to the writings of Nancy Charlotte Roberts of the Naval Postgraduate School, specifically her 2000 article, “Wicked Problems and Network Approaches to Resolution, which was published in the International Public Management Review. Burnside noted that Roberts has defined “wicked” problems as ones that “lack a definitive, standard problem-solving formula; straddle organizational and disciplinary boundaries; involve complex interdependencies, multiple stakeholders, and conflicting agendas; are time-intensive; and are never completely solved.” Burnside said that climate change, nuclear proliferation, and yes, biomedical AI, can easily be described as “wicked problems.”

So, how to handle the “wicked problem of biomedical AI development”? Burnside told her audience that there are three absolutely key elements involved: a commitment to engage stakeholders; analysis based on both quantitative and qualitative techniques; and decision-making continuously aligned among stakeholders and focused on outcomes. Among the key concerns, she noted, being expressed by clinicians and others in healthcare, include the fear of an eventual over-reliance on AI, with humans eventually losing their own analytical abilities; and the fear of the loss of the ability to engage stakeholders, listen to their concerns, discuss those concerns, assess them, and participate in collaborative decision-making.

“What do I believe?” Burnside asked rhetorically. “I believe that we will need effective leadership that is task-relevant” in order to achieve success with AI in clinical settings; and also, that “Successful leaders will adapt their leadership style to the performance readiness of the stakeholders,” in order to engage them in the work of evaluation, assessment, and collaborative decision-making, around AI.

Burnside walked her audience through the results of several recent surveys of radiologists and other physicians, conducted with physicians in the United States, the United Kingdom, and Italy, and said that it’s clear that clinicians are being thoughtful everywhere in terms of thinking through and expressing their concerns. A survey of members of the American College of Radiology (ACR) in April and May of this year found universal consensus in the idea that there will be clinical benefit that will be derived from the leveraging of AI in clinical settings; 0 percent off the respondents to that survey expressed the idea that there would be no benefit. Meanwhile, members of SIRM, the Società Italiana di Radiologia Medica e Interventistica, or Italian Society of Medical and Interventional Radiology, responded in 2019 that they were most focused on error reduction, work optimization, and patient care personalization, and were most worried about potential reputational damage for radiologists. Meanwhile, a survey of primary care physicians here in the US found recently that the PCPs want high sensitivity, high specificity, high radiologist involvement, transparency, diversity, and inclusivity to be among the outcomes of any leveraging of AI tools in radiology.

And importantly, Burnside reported on the results of a survey that she and her colleagues conducted explicitly for the purpose of her lecture, with radiology chairs nationwide surveyed, via SCARD, the Society of Chairs of Academic Radiology Departments. Significantly, 93 percent of those radiology chairs expressed optimism about AI in general, while 86 percent were optimistic about generative AI. The most important outcomes they are looking for from the leveraging of AI are quality and efficiency and a reduction in radiologist burnout. They were less concerned about impact on salaries, cost, education, or equity concerns.

Given all those complex elements, Burnside told her audience that the bottom line, going back to the “wicked problem” frame, will be that the leaders leading everyone forward around the leveraging of AI, especially generative AI, in radiology, will need to work through a host of policy issues, work toward achieving a variety of returns on investment, and hyper-focus on building trust and understanding among all the stakeholders in the processes ahead. And, she added, the following should be considered criteria to evaluate for the implementation of any types of AI: local performance; scientific evidence; fairness, bias, lack of harm; technical readiness and workflow impact; value and cost; and clinical impact.

“Policies really are part of the key,” she said. “And, we need to work diligently on developing understanding,” with the need to find the resources and support to develop data sets, and the identification of known local environments in which the tools can be tested, being important as well. “Leadership is really sitting in your seat!” she told the audience, meaning that they, the audience members must be leaders in this work. “You have an important role to play,” she concluded. “Proudly tackle the tame, while always keeping an eye on the wicked.”

 

Sponsored Recommendations

Enhancing Remote Radiology: How Zero Trust Access Revolutionizes Healthcare Connectivity

This content details how a cloud-enabled zero trust architecture ensures high performance, compliance, and scalability, overcoming the limitations of traditional VPN solutions...

Spotlight on Artificial Intelligence

Unlock the potential of AI in our latest series. Discover how AI is revolutionizing clinical decision support, improving workflow efficiency, and transforming medical documentation...

Beyond the VPN: Zero Trust Access for a Healthcare Hybrid Work Environment

This whitepaper explores how a cloud-enabled zero trust architecture ensures secure, least privileged access to applications, meeting regulatory requirements and enhancing user...

Enhancing Remote Radiology: How Zero Trust Access Revolutionizes Healthcare Connectivity

This content details how a cloud-enabled zero trust architecture ensures high performance, compliance, and scalability, overcoming the limitations of traditional VPN solutions...