Mathematical modelling frameworks are becoming the foundation of in silico studies of the evolutionary dynamics of cell-to-cell communication in bacterial populations living in biofilms or growing in nutrient media. A significant trend in recent decades has been the development of hybrid models for describing complex, difficult to formalize microbial systems to predict and control their states. The present work continues these efforts by extending the hybrid framework into three advanced directions that complement the previously established models. The first direction addresses the development of a reaction-diffusion model that integrates nutrient-dependent bacterial growth, quorum-sensing-mediated population regulation, and dynamic resistance factor production during antibiotic treatment. Simulation results suggest the possibility of propagating wave fronts under certain parameter conditions, and the analysis indicates that the wave velocity and spatial profile may be influenced by the nutrient diffusion coefficient and the quorum activation threshold. The results indicate that the efficacy of antibiotic therapy depends non-monotonically on quorum-sensing intensity. Weak communication fails to trigger the diffusion barrier, whereas excessive signaling undermines bacterial protection through negative feedback. The second direction introduces an advanced hybrid computational framework for the discrete-in-space dynamical modeling of bacterial biofilms. The approach combines a cellular automaton on a hexagonal lattice with discrete analogues of reaction-diffusion equations governing nutrient and signaling molecule distribution, incorporating a quorum sensing feedback mechanism that links local signal concentration to biofilm spreading. A two-parameter analysis reveals a curved transition boundary in the nutrient-threshold plane, demonstrating that the effective quorum sensing activation threshold depends on nutrient availability. The third direction explores the application of physics-informed neural networks for solving direct and inverse problems in quorum-sensing dynamics. For a basic model of bacterial quorum sensing that combines the signaling molecule concentration, the degradation enzyme activity, and the bacterial population size, the PINN is set up to predict the time course of these variables and to recover the parameters governing the influence of natural enzymatic degradation.