A recent study published in the journal Acta Materia Medica has demonstrated the potential of graph neural networks (GNNs) in drug discovery, particularly for complex conditions like osteoarthritis (OA). Traditional Chinese medicine, known for its therapeutic benefits in treating OA, faces challenges in identifying active compounds due to the complexity of herbal compositions. This research successfully employed an in-house GNN model to identify Paederia scandens as a promising treatment candidate. Experimental validation confirmed that this herbal remedy improved cartilage metabolic homeostasis and reduced subchondral bone sclerosis. The study pinpointed asperuloside, a key constituent of Paederia scandens, as the bioactive compound responsible for these effects. Further analysis, including transcriptomic profiling, showed that asperuloside reshapes cartilage gene expression, downregulates inflammatory signaling pathways, and enhances cartilage matrix synthesis while decreasing degradation. Additionally, Integrin Subunit Beta 1 was identified as a potential central regulatory hub in this mechanism. The findings underscore the critical role of GNNs in merging computational predictions with biological validation, paving the way for advancements in precision drug development and the modernization of traditional Chinese medicine.