Abstract
Pain is a complex condition affecting a large portion of the population.Accurate and consistent evaluation is essential for individuals experiencingpain, and it supports the development of effective and advanced managementstrategies. Automatic pain assessment systems provide continuous monitoring andsupport clinical decision-making, aiming to reduce distress and preventfunctional decline. This study has been submitted to the \textit{SecondMultimodal Sensing Grand Challenge for Next-Gen Pain Assessment (AI4PAIN)}. Theproposed method introduces a pipeline that leverages respiration as the inputsignal and incorporates a highly efficient cross-attention transformeralongside a multi-windowing strategy. Extensive experiments demonstrate thatrespiration is a valuable physiological modality for pain assessment. Moreover,experiments revealed that compact and efficient models, when properlyoptimized, can achieve strong performance, often surpassing largercounterparts. The proposed multi-window approach effectively captures bothshort-term and long-term features, as well as global characteristics, therebyenhancing the model's representational capacity.