The role of artificial intelligence in equine colic: A scoping review of diagnostic, prognostic, and decision-support applications

Abstract

Equine colic, strangulating colic in particular, remains one of the leading causes of mortality in horses, with timely diagnosis and accurate prognostic assessment being critical for clinical decision-making. In recent years, artificial intelligence (AI) has been increasingly applied to support diagnostic and prognostic evaluation in veterinary medicine. However, the scope, methodological characteristics, and performance of AI models in equine colic have not been systematically mapped. Therefore, this scoping review was conducted to summarize current AI applications in equine colic, identify commonly used algorithms, describe reported performance metrics, and highlight methodological gaps affecting clinical translation. 16 studies published between 2015 and 2025 were included, comprising 104 AI models applied to equine colic. Most models focused on prognostic prediction, particularly survival outcome prediction. Logistic regression was the most frequently used method, followed by random forest (RF) and other ensemble approaches. Across studies, RF generally demonstrated strong discriminative performance for survival prediction, with reported area under the curve values ranging from 0.79 to 0.99 and accuracy often exceeding 80%. However, external validation, calibration assessment, and standardized preprocessing strategies were infrequently reported. Limited handling of class imbalance and inconsistent reporting practices further reduced reproducibility and generalizability. Findings highlight that AI applications demonstrate promising potential for prognostic prediction and clinical decision-making in equine colic, but important methodological barriers remain. Future research should emphasize multicenter collaboration, external validation, calibration assessment, standardized reporting frameworks, imbalance-management strategies, explainable AI integration, and multimodal monitoring systems to facilitate clinical translation of AI tools in equine practice.