Abstarct: Protein kinases, particularly LIM kinase (LIMK), which plays a crucial role in cytoskeletal regulation and cancer mextastasis, have emerged as important therapeutic targets in modern drug discovery. Considering the significant role of LIMK in cancer cell invasion and migration, the development of selective LIMK inhibitors represents a promising strategy for the discovery of novel anticancer therapies. Tetrahydropyridine derivatives, owing to their structural flexibility and ability to establish diverse molecular interactions, have demonstrated considerable potential as LIMK inhibitors. However, the experimental determination of the inhibition constant (Ki) of these compounds is both time-consuming and costly. In the present study, Quantitative Structure–Activity Relationship (QSAR) modeling combined with molecular docking was employed as an efficient computational approach for the rapid prediction of the inhibitory activity of tetrahydropyridine derivatives and for the identification of potential LIMK inhibitors. The Adaptive Least Absolute Shrinkage and Selection Operator (ALASSO) was utilized for both molecular descxriptor selection and QSAR model development. By selecting the most informative descxriptors and eliminating redundant variables, ALASSO enabled the construction of a sparse, interpretable, and highly predictive QSAR model while reducing the risk of overfitting. Application of the ALASSO algorithm resulted in a final QSAR model containing seven molecular descxriptors. The predictive performance of the model was evaluated using an external test set, leave-one-out (LOO) cross-validation, and several statistical validation criteria. The model achieved an MSE of 0.10 and an R² value of 0.87 for the external test set, while the Q² value obtained from LOO cross-validation confirmed the robustness and predictive capability of the developed model. Furthermore, the resulting sparse model, consisting of only seven descxriptors, demonstrated satisfactory predictive accuracy. The obtained R² and Q² values exceeded the generally accepted threshold values of 0.60 and 0.50, respectively, indicating that the proposed QSAR model is statistically reliable and possesses good predictive ability. The developed QSAR model was subsequently applied to predict the biological activity of newly designed compounds. The predicted compounds were further evaluated by molecular docking, and the docking results supported the reliability of the QSAR predictions. Overall, this study overcomes several limitations of conventional QSAR models and provides an effective computational frxamework for the rational design and optimization of novel tetrahydropyridine derivatives as potent LIMK inhibitors.