A Systematic Review of Personalized and Adaptive Learning Systems: Technologies, Approaches, and Educational Outcomes

Abstract

The integration of machine learning (ML) into adaptive learning platforms has signifi- cantly advanced personalized education in higher education, offering effective solutions to challenges such as student engagement, retention, and dropout prevention. This systematic review, conducted following PRISMA guidelines, synthesizes empirical re search published between 2020 and 2025 to explore how ML-powered adaptive systems personalize learning through real-time behavioral analytics, learning style detection, and dynamic content adaptation. The review highlights key technologies, including multilayer perceptrons, support vector machines, and reinforcement learning, which enable individualized learning pathways tailored to cognitive styles and preferences. Evidence from diverse global contexts, including Sri Lanka, demonstrates improved academic performance, enhanced learner engagement, and reduced dropout rates. However, persistent challenges include scalability, data privacy, algorithmic bias, and the need for human oversight to ensure ethical implementation. The paper concludes by recommending best practices for integrating adaptive systems in higher education and outlines future research directions focusing on long-term impacts, cross-disciplinary applications, and socio-emotional learning integration. This review underscores the transformative potential of ML-enhanced adaptive learning platforms to foster equitable, personalized, and learner-centered education.