Abstract
<jats:p>Pain assessment remains one of the most significant challenges in equine clinical practice and welfare monitoring. Despite advances in behavioural pain scales, facial expression scoring, gait analysis and physiological monitoring, current approaches remain fragmented and largely dependent on subjective interpretation — a limitation that is particularly relevant in horses, where pain expression is often subtle, context-dependent and shaped by species-specific behavioural adaptations. We argue that equine pain should be conceptualised as a multidimensional neurobehavioural state emerging from interactions among nociceptive processing, affective responses, autonomic regulation, motor adaptations and environmental context. Rather than evaluating clinical signs, facial expression, behaviour, movement and autonomic physiology independently, future research should integrate them into multimodal biomarkers capable of objectively characterising pain states. To this end, we introduce the Equine Multimodal Pain Biomarker (EMPB): an integrated biological signature derived from behavioural, physiological and neurophysiological domains that reflects the presence, severity or progression of pain in horses. Within this framework, artificial intelligence is viewed not as a replacement for clinical expertise but as a tool for combining heterogeneous data streams into biologically meaningful, clinically actionable representations of pain. Finally, we propose a five-phase roadmap for the development, validation and clinical translation of objective pain biomarkers in horses. The EMPB framework provides a foundation for future research aimed at improving pain assessment, supporting veterinary decision-making and advancing equine welfare.</jats:p>