Artificial intelligence in oncology practice: National perspectives on adoption, confidence, and implementation readiness.

Abstract

712 Background: Artificial intelligence (AI) is gaining attention in oncology care for its potential to improve operational efficiency, reduce administrative burden, and support clinical workflows. However, concerns about governance, safety, workforce readiness, and clinical integration may hinder adoption. To assess implementation readiness, the Association of Cancer Care Centers (ACCC) conducted a national survey of oncology professionals across care settings. Methods: ACCC, with input from a multidisciplinary advisory committee, developed and distributed a 26-item national survey from May to August 2025. The survey examined AI adoption, confidence in evaluating and using AI, implementation barriers, governance readiness, and priority oncology use cases. Responses from multidisciplinary staff at U.S. cancer programs were analyzed using descriptive and stratified statistics in Stata 18. Open-ended responses were analyzed using rapid inductive thematic analysis. Results: A total of 168 respondents from 36 states completed the survey, including participants from community (46%) and academic/NCI-designated (44%) cancer programs. Respondents represented care delivery (61%) and administrative/operational roles (39%). Across settings, enthusiasm for AI adoption was high, particularly for operational and workflow applications, with 10%-30% of organizations reporting they had implemented or were implementing AI tools. Care delivery respondents reported lower confidence than administrative/operational respondents in describing AI use, discussing benefits, and critically evaluating AI systems, with 26% reporting “not at all confident” versus 13% of admin/ops respondents. Respondents with prior AI experience were more likely to support expanded use of AI for clinical trial matching, prior authorization, patient education, navigation, chatbots, and real-time alerts. Rural respondents and individuals without prior AI experience expressed greater concern about negative impacts on patient-provider communication. Qualitative findings identified recurring implementation priorities, including multidisciplinary stakeholder engagement, workforce education, governance infrastructure, evidence of effectiveness, workflow integration, and organizational oversight. Conclusions: These 2025 findings show strong interest in integrating AI into oncology practice while highlighting limited governance structures and uneven workforce confidence. They also identify opportunities to strengthen AI literacy, implementation readiness, and governance across cancer programs. Targeted educational and operational support focused on evaluation frameworks, multidisciplinary engagement, and responsible implementation may help programs adopt AI in ways that improve efficiency while preserving patient-centered care.