Abstract The primary objective of this study is to enhance the accuracy of reservoir injectivity determination in water injection programs, a vital component of secondary oil recovery. We aim to address the limitations of traditional methods, specifically in the context of thin laminated sands, by introducing a novel multilayer reservoir modeling technique. Our central focus is on improving injectivity characterization in mature fields through the synergistic use of advanced petrophysical logs, Artificial Neural Networks (ANN), and Heterogeneous Rock Analysis. The paper will also showcase the use of machine learning in automatic identification of reservoir layers customized for the injection fall-off pressure transient analysis. Our innovative approach is founded on the fusion of comprehensive petrophysical data and the computational capabilities of ANNs and Heterogeneous Rock Analysis. This methodology combines a spectrum of petrophysical logs, ranging from basic lithology logs to specialized logs like spectroscopy, nuclear magnetic resonance, and elemental analysis, while also integrating borehole image logs. ANNs are employed to facilitate rock typing and the creation of facies models. This framework enables the generation of hydraulic flow units and the derivation of reservoir quality (RQ), which is then merged with completion quality (CQ) to define multiple layers within the reservoir models. Each of these reservoir layers own their unique set of parameters which act as priori for the modeling and regression process in the transient analysis. Our multi-layer modeling approach offers notable operational efficiencies, especially during the injection fall-off pressure transient analysis. When applied to a 3-layer reservoir model, we observed a substantial 30% reduction in modeling and regression time. Beyond the time-saving advantage, this methodology significantly enhances the alignment between empirical data and modeled predictions. Importantly, it provides a marked improvement in the vertical resolution of the injection fall-off analysis, while also reducing the time and effort typically required for manual modeling and regression tasks. In contrast to traditional homogenous modeling, which often yields a single value for permeability and its anisotropy in thin laminated or heterogeneous sands, our approach offers a spectrum of values. This nuanced understanding and improved model fidelity lead to a noticeable enhancement in injection efficiency when these diverse values are incorporated into numerical reservoir models. This paper introduces a groundbreaking integration of petrophysical analysis with injection fall-off transient methodologies, offering a fresh perspective on reservoir injectivity characterization. The fusion of advanced petrophysical logs, ANNs, and Heterogeneous Rock Analysis not only elevates the quality of reservoir models but also refines the vertical resolution of key reservoir characterization parameters, including permeability, skin, and permeability anisotropy. This approach is particularly valuable for those navigating the complexities of heterogeneous sand formations and striving for optimized oil recovery in mature fields.