By Moira Schieke, MD, and Tom Lawry
Fixing big problems takes inspiration. When we consider the current United States health care crisis, the perspective of patients and their families it is more than sufficient inspiration.
Patients today face increasing rates of medical errors, report poor experiences in interacting with their health care system, and bear the ultimate burden from the world’s most expensive underperforming health care system. According to groups such as the Institutes of Medicine and corroborated by others, the rate of medical errors in the US is roughly 10% to 20%.1 A staggering 67% of patients report poor experiences. The cost of the US health care system continues to escalate, faster than any other country, rising from approximately $1T in 1996 to approximately $4T in 2022.2
Who wins? Increasing costs correlate with increasing revenues by large corporate vendors, including IT companies, who reap billions.3 Who pays the bill? US taxpayers, employers, company-sponsored health plans, patients, and their families are all struggling with rising costs.3,4,5,6,7 In fact, medical bills are the number one cause of bankruptcy in the United States. Who loses? Everyone. Today, one-third of health care workers agree with the statement, “the American health care system is on the verge of collapse.”6 Forty-seven percent say they are likely to leave the field by 2025.6 US population health statistics show the US has the highest death rate for avoidable or treatable conditions, the highest maternal and infant mortality, and among the highest suicide rates of any high-income nation.2
Medical professionals work tirelessly to benefit patients’ lives, improve their experiences, and better the nation’s health, but every day they are confronted with endless trade-offs. Everyday realities undermine the ability of many of the best trained medical professionals in the world to perform as they wish for their patients. Radiologists are pushed to throughput increasingly high volumes of reports to generate revenues to cover escalating operational expenses. This comes at the expense of quality decision making and physician well-being. These tradeoffs are growing daily, as clinicians are forced to manage increasingly large precision-era datasets, which only adds to their decision complexity in fast-paced, real-time clinical care environments.
A Paradigm Shift
Fixing the US health care system will take successful digital transformation. Medical journals, outlets, and social media have been ablaze for years about digital transformation to fix the problems that ail us. Yet, are we really on the path to success?
In radiology, we have seen a rapid uptake of cloud and AI technology. We do not argue that this is progress, as we have seen a few successful use cases. Yet, AI and cloud have been mostly used for either dangerous efforts to supplant expert decision making (eg, “decision automation”) or incremental advances on existing paradigms, such as decades-old “decision support.” What do these terms means? See figure 1. 8,9
In decision automation, the human or AI asks the questions and AI makes decisions, with or without human-in-the-loop assistance. Predictably, AI autonomous systems in clinical environments have led to dangerous patient care outcomes, as documented by the ACR and FDA.4 In contrast, decision support allows the human to ask questions and makes decisions, and the computer assists by providing more data or information.
Decision support remains the pervasive decades-old paradigm of today. With very few exceptions, all data and information is harvested and viewed by a physician, such as a radiologist, who then makes the final decision. This paradigm will always be needed in chaotic decisions spaces which require our unique human capabilities and where AI systems cannot safely function independently. AI point solutions aren’t reliable in clinical environments, generating approximately 93% brittle failure rates.1 Further, innumerable AI point solutions on single types of data generate more data and analysis for viewing and use by human decision makers, thus adding to their physical and mental strain. AI automation and other approaches such as radiomics and integrated diagnostics are not providing the leap beyond our decades-old “decision support” paradigm that’s needed.
So, what’s the paradigm switch? It’s a middle ground called “decision augmentation.” It can be successfully enabled in situations where there is nonchaotic decision complexity and where there are ways we can use AI for decision components that are repeated and predictable. In “decision augmentation,” humans ask the questions, both the computers and the humans make decisions, and the final decision is made by the human.
In order to manage today’s massive precision-era datasets, we need a paradigm switch to decision augmentation. This concept is at the heart of assuring a successful digital transformation, which is still lacking today.
Complex Human Decisions
Personalized and precision medicine has been a goal for many years, as its potential to radically improve our health care system is clear. With the right vision, we have made enormous progress—from genetics, proteomics, and liquid biopsies to burgeoning new types of noninvasive imaging.
However, by only focusing on collecting data without also considering how those data could be managed within high pressure clinical environments, we have created a whole new set of problems. Exploding volumes of data held in 1980s siloes force clinical decision making to engage in enormous amounts of wasteful toil to gather and organize all needed data for each decision. Further, we are generating clinical decision errors due to cognitive overload and forcing enormous visual multitasking. Fortunately, we have solutions.
“Visual analytics” and “decision intelligence” (DI)are two core disciplines which have yet to be introduced to health care in any significant way. Yet, these fields are fundamental for facilitating expert decision augmentation in complex decision spaces.
Health care’s problem of increasing decision complexity and associated errors in fast-paced environments is not unique. In fact, this problem has been blamed for the US’s failure to see clear intel leading up to the 9-11 terrorist attacks. In 2017, in recognition of these fatal failings, the US Department of Homeland Security initiated a high-level think tank of experts, including top cognitive scientists, around the new academic field of visual analytics.5 Visual analytics is an outgrowth of information visualization and scientific visualization, which focuses on analytical reasoning facilitated by interactive visual interfaces.
DI is another field that is critical when designing systems for decision augmentation. DI is a methodology that analyzes complex decision pathways to define the components where decisions can be safely handed off to computers. Importantly, unlike AI automation point solutions, decision intelligence assures synthesized computer-human decision making is always mapped to desired outcomes, an important safety lever for patient care. It creates a framework that can be continuously monitored and improved. Revolutionary DI systems have been validated in domains such as mortgage approvals, Formula 1 racing, and Google Maps route optimization.10
A Rapidly Accelerating Future
Successful digital transformation is about capturing new value at scale, which requires us to embrace change.6 Today, we have widely accessible AI and cloud technology that can be leveraged to design decision augmentation systems to facilitate this new value at scale. Decision augmentation offers us the safest route to progress, where humans ask questions, computers and humans make decisions, and humans make the final decisions. This is the brave new world we are entering, and many believe it’s moving faster than it should. Crucially, this inevitable change can’t happen safely and effectively within the field of medicine without physician leadership.
New cloud technology, such as data meshes and data fabrics, will allow us to modernize our informatics systems to create the intuitive order that’s needed. Designing these new informatics systems requires expert medical, professional-subject-matter expertise for every potential medical use case. Considering that today’s health care data are 80% unstructured and medical imaging represents 90% of the digital health footprint, radiologists will be uniquely important as subject matter experts for modernizing informatics systems leveraging new cloud capabilities.1
Achieving this bold vision of the future will require technological and cultural changes that may seem daunting. Yet, fueled by the magnitude of our health care crisis, there is simply no longer a question of whether a future of AI and cloud will happen. It’s now more a question of who will manage it.
For radiology, though, isn’t this just more history repeating?
Successful transformation takes “gas and guardrails.”6 Radiology has been at the forefront of technological advancement in health care for over a century. The field has shepherded many technology shifts, with X-ray, CT, MRI, and nuclear medicine, as well as being the first field to digitize in the 1980s with the advent of PACS. While AI and cloud present big new challenges, in the big picture, radiology has evolved the guardrails to guide new technological innovation in medicine over its amazing 100-year history. Under radiology’s leadership, medicine has been gracefully transformed, for the better, by technological advances that were embraced for better patient care.
Radiology faces a leadership imperative.
As we continue to work toward our vision of personalized and precision-era care, it’s time to recognize the need to change course and embrace a paradigm shift. We must guide a successful digital transformation to realize new value at scale. Medicine has generated new problems with precision-era datasets that we must now solve by embracing the solutions provided by new, quickly accelerating technology. A metamorphosis requires clinical leaders and other stakeholders to think differently. They must accept a culture of innovation and technological advancement, which they hadn’t before imagined in the realm of health care. Foremost thought leaders for leadership strategy recognize that the paradigms that worked for centuries are no longer adequate for today’s rapidly changing world. Medicine’s traditional hierarchies threaten success. Leadership in today’s world is about aligning with mission and vision and providing the right conditions to harness the “collective genius of people.”7
Good leadership will be a primary driver for success. It will take hard work and responsible guidance.
We all understand what success could mean for everyone. What could monitored and continuously improved patient care outcomes do for population health? What could a revitalization of the patient experience mean for patients’ and their families’ emotional and physical well-being? What could it mean for patients’ and their families’ financial health? What will the impact be on the United States and the world? Once we know we are on the right path, imagine how much more rewarding it will feel, once again, to be a health care professional.
We hope you are inspired to help lead the way.
— Moira Schieke, MD, is the founder and CEO of Cubismi, Inc, a digital medicine innovator and pioneer, a cancer imaging and MRI expert, and a board-certified clinical radiologist. She is an advocate for patient digital rights and breaking down legacy barriers to improved patient care.
— Tom Lawry is a strategic advisor to health leaders worldwide and best-selling author of Hacking Healthcare: How AI and the Intelligence Revolution will Reboot an Ailing System. During his more than 30-year career, he has served in the C-suite of health care systems, been a successful startup founder, and served as the national director for AI—Health and Life Sciences at Microsoft.
1. Moira Schieke. Workflow is for Machines. Decision Flow is for Humans. Medium website. https://medium.com/the-cubismi-blog/workflow-is-for-machines-decision-flow-is-for-humans-bce706c5dd79. Published December 20, 2022.
2. US Health Care from a Global Perspective, 2022: Accelerating Spending, Worsening Outcomes. The Commonwealth Fund website. https://www.commonwealthfund.org/publications/issue-briefs/2023/jan/us-health-care-global-perspective-2022. Published January 31, 2023.
3. Shapiro, Martin F. The Present Illness: American Health Care and Its Afflictions. Baltimore: Johns Hopkins University Press; 2023.
4. American College of Radiology. “Public Workshop – Evolving Role of Artificial intelligence in Radiological Imaging;” Comments of the American College of Radiology. https://www.acr.org/-/media/ACR/Files/Advocacy/Regulatory-Issues/acr_rsna_comments_fda-ai-evolvingrole-ws_6-30-2020.pdf. Published 2020.
5. National Visualization and Analytics Center. Illuminating the Path: The Research and Development Agenda for Visual Analytics. https://ils.unc.edu/courses/2017_fall/inls641_001/books/RD_Agenda_VisualAnalytics.pdf. Published 2005.
6. Tom Lawry. Hacking Healthcare: How AI and the Intelligence Revolution Will Reboot an Ailing System: Routledge. Taylor & Francis Group, A Productivity Press Book: 376 pages; hard copy. ISBN: 988-1-032-26016-7.
7. Allen RK. The Leadership Imperative. The Center for Organizational Design website. https://centerod.com/2014/07/leadership-imperative/
8. Using AI in Decision Making: When and Why. Gartner website. https://www.gartner.com/smarterwithgartner/would-you-let-artificial-intelligence-make-your-pay-decisions. Published June 2, 2021.
9. 7 Levels of Hybrid Human and AI Decision Making. Gartner website. https://www.gartner.com/en/documents/4003141. Published June 30, 2021.
10. Decision Intelligence Is the Near Future of Decision Making. Gartner website. https://www.gartner.com/en/documents/4004300. Published August 3, 2021.