<p>The article introduces the entropy regularized shrinkage estimator (ERSE) method, a novel approach to the estimation of parameters in linear mixed models through the application of a penalty method inspired by information-theoretic regularization techniques. The ERSE is based on the application of the ridge penalty to the loss function, with the quantity and dispersion of the coefficients being simultaneously modified through the incorporation of the entropy term. The entropy term serves the purpose of stabilizing the estimate in the event of the presence of heterogeneous noise, a limited sample size, and the occurrence of multicollinearity. The optimization of the ERSE loss function is demonstrated to be achievable through the derivation of the gradient from the theoretical analysis, with the optimization being demonstrated to be differentiable and convex. The performance of the proposed method is demonstrated through the execution of simulation tests with varying sample sizes and levels of noise, allowing the comparison of the performance of the proposed method with existing benchmark methods such as the Elastic Net, Generalized Least Squares, Ridge, and Lasso methods. The results show that the proposed method is characterized by a constant entropy behavior, with the estimate bias being reduced and the mean squared error being minimized. The proposed method is also accompanied by a Shiny-based implementation tool for the visualization of the performance of the proposed method under various parameter configurations, helping the reader understand the proposed method in a real-world scenario.</p>