Behavioral assessment of pain in rodents: advances from evoked responses to spontaneous states and multimodal approaches

Pain is widely recognized as a leading global health problem that markedly diminishes quality of life. Although assessment lies at the core of pain medicine, robust quantification remains difficult. In preclinical research, commonly used behavioral assays often blur the distinction between spontaneous pain and stimulus-evoked responses. Here, we review recent advances, clarify the conceptual and operational boundaries between spontaneous and evoked pain, and provide a multidimensional comparison of major traditional behavioral paradigms. To address shortcomings in objectivity and reproducibility, we also summarize emerging evaluation strategies. Finally, leveraging bioinformatics and machine learning, we identify pain-associated metrics and propose building multimodal datasets and AI-driven feature-extraction pipelines to enhance the translational value of animal data for clinical pain research.