Deep Learning and Reinforcement Learning for Assessing and Enhancing Academic Performance in University Students: A Scoping Review

University students often face challenges in managing academic demands and difficulties like time management, task prioritization, and effective study strategies. This scoping review investigates the application of Deep Learning (DL) and Reinforcement Learning (RL) in evaluating and enhancing academic performance, focusing on their practical applications, limitations, and future potential. Using PRISMA guidelines, 27 empirical studies published between 2014 and 2024 were analyzed. These studies utilized advanced DL and RL technologies, including neural networks and adaptive algorithms, to support personalized learning and performance prediction across diverse university contexts. Key findings highlight DL’s ability to accurately predict academic outcomes and identify at-risk students, with models achieving high accuracy in areas like dropout prediction and language proficiency assessments. RL proved effective in optimizing learning pathways and tailoring interventions, dynamically adapting to individual student needs. The review emphasizes significant improvements in grades, engagement, and learning efficiency enabled by AI-driven systems. However, challenges persist, including scalability, resource demands, and the need for transparent and interpretable models. Future research could focus on diverse datasets, multimodal inputs, and long-term evaluations to enhance the applicability of these technologies. By integrating DL and RL, higher education can foster personalized, adaptive learning environments, improving academic outcomes and inclusivity.