As radiology departments and practices around the world begin to adopt Artificial Intelligence (AI) technologies, radiologists must take the lead in determining how AI is applied throughout the imaging life cycle. That includes assessing how data is captured and utilized to build the AI tool.
“It’s our job to lead the way through the clinical adoption of AI, even when it comes to the nitty gritty details of looking at the data,” said K. Elizabeth Hawk, MS, MD, PhD, Interim Chief of Nuclear Medicine, University of California San Diego. While looking at the data might seem irrelevant to radiologists, Dr. Hawk said it’s essential to doing what they do best – caring for the patient.
“As we begin to look at the data, we need to think about the impact on patient care. We know AI can be successfully applied in a variety of different practice settings, but we need to stay rooted in the clinical questions,” she said. “It’s our job to look at these tools and make sure we’re responsibly using them and still handling patient care with a tremendous sense of responsibility and respect.”
The Ethical Implications of AI
New technology, especially disruptive technology like AI, has the potential to either improve or worsen healthcare inequities. This is why Dr. Hawk is particularly interested in the ethical implications of AI for radiologists who, like all clinicians, are guided by the Hippocratic Oath to “do no harm.”
She notes that gathering data from a homogeneous patient population teaches the AI tool to serve only that population. If that tool is then clinically applied across other heterogeneous populations or underserved communities, it won’t work as well.
“We need to be responsible for which data we’re using when we create these new algorithms and making sure that, as a physician with the patient at the center of our work, we’re creating a disruptive technology that doesn’t deepen healthcare inequalities across our global community,” she said.
The Need for Diversity of Data
To avoid exacerbating healthcare inequities, the best AI tools will be built with data integrity, which is the key to combating fear and improving adaptation across a wide range of patients. “Diversity is tremendously important across the imaging sciences, particularly in informatics and AI. That’s something we need to be cognizant of as we continue to grow and mentor leaders in this field,” Dr. Hawk said.
Diversity of data will lead to stronger AI tools with the power to improve the clinician and patient experience. “The best AI will build trust, improve patient care and ultimately help us as physicians deliver a higher level of clinical care to a heterogeneous and wonderfully diverse patient population,” she explained.
The ultimate goal is finding a balance between ethical AI tools that are widely applicable, yet still allow clinicians to meet every patient’s unique needs. Dr. Hawk said this will have the twin benefits of furthering both standardization and personalization of care.
“I like the concept of creating tools that can be diversely and widely applied across large practices, yet that still allow individualization for patient care,” she said. “We need to find the balance in creating a tool that elevates the standard of care, but also gives the physician the flexibility to individualize that precision medicine for the patient.”
The Design and Implementation Process
No matter what AI tools are built, adopted, and implemented, Dr. Hawk stressed that radiologists are the key to making sure these new technologies work for patients and clinicians. “We need to capture the right data with the right people. We need to always understand the impact on patient care and ask if we’re developing a disruptive technology that's ultimately going to lessen healthcare inequity. And then, we really need to manage the process together through diverse, thoughtful leadership to achieve a transformative innovation that's going to change the art of medicine that, although new and different, stays rooted in our Hippocratic Oath.”
Back To Top
AI Implementation | The Right Data Includes Diverse Data . Appl Radiol.
McKenna Bryant is a freelance healthcare writer based in Nashotah, WI.