Periodontal disease, or periodontitis, is a chronic inflammatory condition affecting the tissues supporting teeth, with epigenetic mechanisms such as DNA methylation, histone modifications, and RNA molecules playing a crucial role in its progression. Histone deacetylase (HDAC) inhibitors have shown potential in treating inflammatory diseases by modulating gene expression to suppress inflammation and promote tissue regeneration. Machine learning models, particularly Kolmogorov-Arnold Networks (KANs), provide an advanced solution for predicting drug-gene associations, offering superior accuracy, efficiency, interpretability, and scalability compared to traditional Multi-Layer Perceptrons (MLPs). This study explores the application of KANs in predicting drug-gene associations of HDAC1 inhibitors for periodontitis. A dataset comprising 533 compounds and genes was analyzed using Cytoscape for network-based topological and functional insights, followed by predictive modeling using KANs. The resulting network contained 326 nodes and 3734 edges, with an average of 23.411 neighbors, a diameter 5, and a characteristic path length of 2.383. The predictive model achieved an impressive accuracy of 96.49 %, as indicated by the F1 score, reflecting balanced performance in classification. The study highlights the potential of KANs in drug discovery for periodontal disease by efficiently predicting drug-gene associations, enabling better understanding and research for experimental validation and clinical applications.