Technology-based learning called blended learning began its revolution with the immediate emergence of full-fledged Internet service-providing systems globally. It’s a renovated idea of integrating a traditional education system combined with e-learning. This hybrid learning provides educational supplementary for traditional classroom-based teachings online. The elementary objective of this review is to devise a new technique called proposed Quasi Recurrent Neural Network_Adam Wild Horse Optimization (QRNN_AWHO) for academic performance prediction in a blended learning model. For that, initially, academic data from a blended learning environment is considered as an input and then data normalization is conducted by employing Z-score normalization. After that, feature selection is performed using mutual information, and finally, academic performance prediction is done by employing Quasi Recurrent Neural Network (QRNN), which is trained using the proposed Adam Wild Horse Optimization (AWHO). Here, AWHO is devised by the amalgamation of Adam Optimizer and Wild Horse Optimization (WHO). The QRNN_AWHO endured to be a precise model in students’ academic performance prediction with an imminent highest performance score for the following metrics, like precision, recall, F-measure, and accuracy of 87.433%, 90.565%, 88.971%, and 85.234% in accordance with the integration of blended learning.