Overexpression of the spleen tyrosine kinase (SYK) has been found associated with different cancer types. Despite the investigation of inhibitors of SYK including fostamatinib, entospletinib, cerdulatinib, and TAK-659 for cancer therapy, their lack of specificity and potential off-target effects remain significant concerns. Addressing the need for targeted and non-toxic SYK inhibitors, this study integrates machine learning with structure-based drug design. Using bioactivity data, we employed machine learning algorithms random forest, to screen an FDA-approved drug library. Molecular docking and dynamics simulations were then conducted to assess binding affinities and stability of identified compounds. Rifabutin, darunavir, and sildenafil were found as promising SYK inhibitors, showing strong interactions and stable conformations. Analysis of RMSD, RMSF, RoG, hydrogen bonding, PCA, and MMGBSA/MM-PBSA supported their efficacy as safer alternatives to current inhibitors. Our findings underscore the value of computational methods in drug discovery and advocate for further experimental validation of these compounds as SYK-targeted therapies. This study aims to advance the development of more effective and safer treatments for cancers associated with SYK overexpression.