Abstarct: Protein kinases, particularly LIM kinase (LIMK), are considered important targets in anticancer drug discovery and development because of their critical roles in regulating the cytoskeleton, as well as cancer cell migration and invasion. Accordingly, tetrahydropyrido-pyrrolopyrimidine derivatives, owing to their structural characteristics and potential to establish favourable interactions with the active site of the protein, may represent promising candidates for LIMK inhibition. However, the experimental determination of the inhibition constant (Ki) for a large number of compounds is a time-consuming and costly process. Therefore, in the present study, quantitative structure–activity relationship (QSAR) modelling was employed to address this challenge. To this end, the Smoothly Clipped Absolute Deviation (SCAD) penalised regression method was applied for variable selection, enabling the identification of the most relevant descxriptors from the initial set of molecular descxriptors. The application of SCAD reduced the number of input variables, controlled model complexity, and facilitated the development of a sparse and interpretable model. Subsequently, the selected descxriptors were used as inputs to an artificial neural network (ANN), and a nonlinear SCAD–ANN model was developed. The performance of the model was evaluated using an external test set, leave-one-out cross-validation (LOO-CV), and various statistical parameters. The results demonstrated that the developed model exhibited satisfactory predictive performance and generalisation ability, with MSE and R² values of 0.03 and 0.91, respectively, for the external test set, while the Q² value obtained from LOO-CV was 0.78. Moreover, the R² and Q² values exceeded the commonly accepted thresholds of 0.60 and 0.50, respectively, in QSAR studies, indicating the satisfactory predictive capability of the developed model. The final model employed only seven molecular descxriptors, thereby providing a sparse model while maintaining good predictive performance. The developed QSAR model was subsequently used to predict the biological activity of proposed compounds. Molecular docking studies were then performed to further investigate the ability of these compounds to interact with the biological target. The docking results indicated favourable interactions between the selected compounds and the active site of LIMK, providing computational evidence consistent with the QSAR predictions for the identification of potential LIMK inhibitors. :