Multimodal and Data-Driven Approaches in Acupuncture Research: Methods, Applications, and Challenges

Acupuncture research increasingly involves heterogeneous and multimodal data that are difficult to analyze using conventional methods. This review summarizes data-driven approaches in acupuncture research within a framework encompassing intervention, response, and contextual data. We discuss causal inference, artificial intelligence, text mining, and integrative analysis, along with their applications in efficacy evaluation, outcome prediction, mechanistic investigation, and clinical decision support. These approaches shift the focus of acupuncture research from population-level average effects toward individualized clinical decision-making by enabling the analysis of treatment heterogeneity and underlying mechanisms. However, current research remains limited by inadequate data standardization, insufficient external validation, and limited model interpretability. Despite these challenges, data-driven approaches offer substantial promise for advancing more rigorous and personalized acupuncture research. Graphical Abstract: http://links.lww.com/AHM/A227