<p>The prediction of crack propagation rates in 2024-T351 aluminum alloy is a critical concern for the aerospace and automotive industries. In this research study, experimental results and machine learning approaches are combined to investigate the effect of grain boundaries on the crack propagation behavior. Mechanical tests on notched specimens and microstructural investigations show that grain boundaries act as physical barriers, effectively slowing crack propagation. Based upon these observations, two prediction protocols are developed using multiple machine learning algorithms. A cubic spline interpolation technique was also applied to generate an expanded dataset from the original experimental data, composed of only 23 experimental data points. Cracks often initiate at notches, while grain boundaries significantly affect crack growth by retarding the propagation rate. The models achieve <i>R</i>² values up to 0.7143 for the CNN and 0.9958 for the MLP. The integrated experimental-machine learning framework provides an effective approach for modeling the fatigue crack propagation of 2024-T351 aluminum alloy.</p>