Adopting Artificial Intelligence in Veterinary Diagnostics: A Scoping Review of Key Challenges

Although artificial intelligence (AI) is increasingly applied to veterinary diagnostics, itsreal-world implementation faces a unique set of challenges. Unlike in human healthcare whereAI systems are developed under structured regulations and validated using large, standardizeddatasets, veterinary medicine operates with less regulatory oversight, more heterogeneous data,and fewer resources. These conditions hinder the development of trustworthy, generalizable, andethically sound AI tools.This scoping review critically examines the current landscape of AI in veterinarydiagnostics, with a focus on the conceptual and technical challenges that impact its clinicaladoption. To map the landscape and refine our search strategy, we initially conducted a manualexploratory search, followed by a systematic search in PubMed Central, which identified studiespublished as early as 2013 through June 2025. After screening using AI-assisted methodsfollowed by manual review, sixty-three articles met the inclusion criteria and were used for thisreview.Our analysis categorized challenges in adopting AI in veterinary diagnostics into threegroups: data acquisition, model development, and implementation. Issues such as limited datawere explicitly acknowledged by most articles, while lack of transparency and externalvalidation were identified as the most common potential limitations. We also highlight possiblesolutions to support more responsible and effective clinical adoption.