AI adoption and workflow optimization following orchestration platform implementation and structured change management

Abstract

Artificial intelligence (AI) adoption in radiology remains limited by workflow integration challenges, lack of interoperability, and barriers to user engagement. We implemented a dual-track socio-technical strategy in a tertiary-care neuroradiology department, combining a vendor-neutral AI orchestration platform with a structured change management program, including user education, workflow standardization, documentation, feedback mechanisms, and governance. In an observational pre–post implementation study, AI adoption was evaluated by comparing a baseline period using a single-vendor solution (September–November 2022) with a post-implementation period following platform deployment (May–July 2025). Adoption was measured using the user-verification rate, defined as the proportion of AI outputs actively reviewed and either accepted or rejected by radiologists. The overall user-verification rate increased from 13.9% (318/2287) at baseline to 61.1% (1778/2910) after implementation, while full verification of examinations containing multiple AI tasks increased to 62.6%. Adoption varied among individual users, and trainees demonstrated higher verification rates when supervised by consultants with high AI adoption. During the post-implementation period, the platform processed 9480 AI tasks across 12 applications, demonstrating scalability across diverse clinical workflows. In conclusion, a combined orchestration platform and change management strategy was associated with substantially increased AI adoption and integration into routine neuroradiology workflow.