Exploring the mechanism of Dieda Qili Tablet on fracture healing based on network pharmacology combined with machine learning models

Dieda Qili Tablet (DDQLT) is a traditional Chinese medicine formulation clinically used for fracture treatment; however, its underlying pharmacological mechanisms remain elusive due to its complex multi-component nature. Network pharmacology was employed to identify active ingredients and potential targets of DDQLT. Intersection targets with fracture-related genes were subjected to protein-protein interaction (PPI) network analysis, GO and KEGG enrichment analyses. Transcriptomic validation was performed using the GSE93215 dataset, followed by machine learning (LASSO, random forest, SVM) to identify hub targets. Molecular docking and 100 ns molecular dynamics (MD) simulations, complemented by MM/PBSA binding free energy calculations and free energy landscape analysis, were conducted to validate target-ligand interactions. We identified 127 active ingredients and 780 DDQLT-related targets, with Myrrha contributing most abundantly. Intersection with 6,444 fracture-related genes yielded 450 common targets, from which 15 core targets were identified via PPI topology. KEGG analysis highlighted JAK-STAT as the second most significantly enriched pathway. Transcriptomic analysis of GSE93215 revealed 296 DEGs in fracture patients, with 6 core targets differentially expressed. Machine learning consistently identified JAK2 as the hub target. Molecular docking demonstrated favorable binding affinities (range: -6.76 to -8.60 kcal/mol) between eight DDQLT ingredients and JAK2, with SD4 exhibiting the optimal binding profile. MD simulations confirmed stable JAK2-SD4 complex formation, with RMSD stabilizing at 0.22–0.30 nm and ligand RMSD remaining below 0.08 nm. MM/PBSA calculations yielded a total binding free energy of -21.90 kcal/mol, with van der Waals interactions serving as the primary driving force (ΔEvdw = -29.21 kcal/mol). Free energy landscape revealed a single, well-defined low-energy basin, indicating specific high-affinity binding. This study systematically elucidated the multi-component, multi-target mechanisms of DDQLT in fracture treatment and identified JAK2 as a central hub target with stable binding characteristics, providing a molecular foundation for understanding its clinical efficacy and highlighting JAK2 as a promising candidate for therapeutic intervention.