Artificial intelligence knowledge, use, and perceived barriers among health care professionals in rare hematological disease networks

Abstract Background AI techniques are increasingly demonstrating potential in improving patient outcomes, with applications rapidly developing in rare hematological diseases (RHDs), including hemoglobinopathies. To guide AI introduction, it is essential to explore healthcare professionals’ (HCPs) knowledge of basic AI concepts, attitudes, and perspectives toward its integration into hematological clinical research and practice. Methods A cross-sectional survey covering demographics, work environment, AI terminology confidence, research involvement, perceived implementation barriers, concerns, and AI’s impact across care phases was conducted among the European hematology community through the networks European Reference Network-EuroBloodNet and HELIOS initiatives and disseminated through convenience sampling via mailing lists and social media (January 8th–June 1st 2025) on the EU Survey Platform. Answers were analyzed through descriptive and comparative methods. Results 73 HCPs from 25 countries, mostly from southern Europe, answered to the survey, with a low approximate participation of 10%. Most respondents (74%) worked in teaching hospitals; 35.6% in research and 64.4% in clinical practice or both. Physicians constituted 49.3% (69.4% senior); other HCPs, including biologists and nurses, made up 50.7%. 74% correctly identified “Machine Learning,“, while for “Deep Learning” only 57.5% chose correctly, with 37% of confident respondents failing. Although 56.2% felt familiar with “Big-Data,” only 31.5% identified its three core characteristics. Overall, 71.2% used at least one AI tool, and 94.6% believed AI would benefit patients. The most cited barriers were lack of infrastructure, personal liability, AI trustworthiness and patient privacy. Conclusion Despite widespread optimism and AI adoption, significant knowledge gaps persist among RHD professionals, who sometimes overestimate their understanding. These findings will inform the design of targeted, multidisciplinary training tailored to different HCP categories, addressing the identified gaps.