CHICAGO -- The hematology/oncology workforce endorses the incorporation of AI into medical education, according to findings presented at ASCO Annual Meeting.
The hem/onc workforce also supports using AI in clinical practice, but concerns about potential challenges and risks persist, survey results showed.
"AI tools are already being used, they're becoming more widely available and they aren't going away," researcher Guilherme Sacchi de Camargo Correia, MD, third-year hematology/oncology fellow at Mayo Clinic in Jacksonville, Florida, told Healio. "They'll be used by physicians, patients and caregivers.
"For that reason, we wanted to understand how the hem/onc workforce understands and perceives this technology and its potential role in their daily lives," he added. "These findings are thought-provoking, and I think they help us get a sense of what we need to think about as we try to move the needle forward."
The evolution of AI and the development of more user-friendly software have contributed to its rapid adoption in health care.
Correia and colleagues aimed to assess perceptions and concerns among the hem/onc workforce about the role of AI in medical education and clinical practice in hopes of gaining insights that would be relevant for integration of AI into the specialty.
Researchers administered anonymous online questionnaires to the full hem/onc workforce at Mayo Clinic's three main sites between Nov. 7, 2024, and Jan. 20, 2025.
The 344 voluntary survey respondents included physicians (faculty, n = 77; fellows, n = 41) and non-physicians (advanced practice providers, n = 49; registered nurses, n = 177).
Two-thirds of respondents (67%) reported having little to no knowledge about AI. A majority (64%) reported having used AI, but more had done so for personal use (51%) than as part of medical education (31%) or clinical practice (28%).
Nearly all participants (94%) indicated they believe AI will be incorporated into hem/onc medical education, and 90% viewed that integration as beneficial.
A majority of physicians -- both faculty (92%) and fellows (85%) -- responded that fellowship programs should integrate AI training into the curriculum.
Most respondents indicated AI could be integrated into medical education by developing materials for patient education (81%), helping educators elaborate on lectures or other materials (71%), developing simulation models (70%) and helping with self-study (70%).
Most respondents (95%) indicated that they think AI will be integrated into hem/onc clinical practice, and 90% indicated they would embrace AI in this setting.
A higher percentage of respondents indicated they believe AI will help with documentation during or after visits (80%) or assist with chart review (73%) than help with analysis of lab results (66%) or help with drug dosing and interaction monitoring (58%).
Most respondents indicated they felt AI will be used to collect patient data for database research (85%), help screen for clinical trial enrollment (78%) and assist with communication with patients following visits (75%). Only half indicated it will be used to generate research hypotheses (51%).
Despite AI's potential utility in clinical practice, 62% of respondents -- including 68% of APPs and registered nurses, and 50.8% of physicians -- expressed concerns about the risks that it may pose.
The most common perceived risks included the potential for inaccuracies or reduced care quality, as well as the possibility that AI would reduce time spent with patients.
In contrast, one-third (33%) reported feeling AI would improve efficiency and increase care quality, and 30% indicated they thought AI would help increase time clinicians spend with patients.
About one in six respondents (14%) expressed worry that their role could be replaced by AI.
"Informally, we hear a lot of people say, 'It's all going to be AI in the future,'" Correia said. "I was pleasantly surprised at the relatively low numbers of respondents who believe AI will, in fact, replace us, because I don't think that will be the reality, either."
The findings suggest proper "systematic education" of the hem/onc workforce is needed to mitigate what Correia and colleagues characterized as "legitimate existing concerns."
"Education is essential, and the importance of that cannot be overstated," Correia said. "We should be thinking about this for fellowship, residency and medical school -- not only for physicians, but for nursing school, APP school and for other allied professionals."
Correia also emphasized the importance of the health care community partnering with AI developers to ensure appropriate and effective integration.
"I don't know how to create these tools, and an AI developer may not know how I want to use AI," Correia said. "This collaboration is key. Right now, it's almost like we are trying to fit a square block into a circle. Sometimes it might fit, but most likely not perfectly. We'd prefer to have a circular block fitting into a circle, and the only way to do that is start at the beginning by having health care professionals participate on the development side."