Hydroxyapatite (HAP) scaffolds created through biological three-dimensional (3D) printing have been extensively utilized in bone tissue engineering. However, clinical outcomes remain unsatisfactory. Stress-induced fractures are a primary cause of failure in clinical bone implant repairs. Finite element analysis (FEA) can predict stress distribution within scaffolds under load, aiding in the identification of stress concentrations. Nevertheless, limitations in the printing fidelity of bio-inks lead to discrepancies between the printed scaffold structure and the theoretical design, resulting in inaccurate predictions. This study introduced an effective approach for predicting the mechanical properties and internal stress distribution of printed scaffolds. HAP scaffolds with varying filling angles (90º, 60º, 45º) were constructed through 3D printing. These scaffolds were analyzed using a combination of optical coherence tomography (P-OCT) scanning and micro-computed tomography (micro-CT) imaging, integrated with FEA, to enhance the accuracy of stress distribution predictions. The findings demonstrated that the mechanical strength of scaffolds predicted by the reconstructed model was closer to experimental values compared to predictions made by commonly used theoretical models (improved from 55% to 78%). During actual testing, stress concentration sites predicted by the reconstructed model were the first to exhibit fragmentation, validating the accuracy of stress distribution predictions. Moreover, scaffolds printed at different infill angles displayed varying degrees of distortion. Scaffolds printed at a 90º angle showed the highest fidelity with the fewest defects, and their stress distributions under different conditions were well-correlated.  The method proposed in this study facilitates more accurate prediction and evaluation of scaffold performance early in development, enabling the selection of scaffolds with suitable mechanical properties, thereby reducing testing cycles and improving safety in bone defect repair.