A systematic review (2010–2024) of smart textbooks and intelligent tutoring systems: Models, methods, subjects, and geographies

Abstract

To synthesize peer-reviewed studies from 2010–2024 on smart textbooks and intelligent tutoring systems, mapping the evolution of models and methods, subject-area coverage, and geographic trends, and cataloging reported advantages and limitations. A PRIS-MA-guided systematic review was conducted across Google Scholar, Scopus, and Web of Science using search terms centered on “smart/intelligent textbook” and “intelli-gent/smart tutor”; from 233 records, 26 studies met inclusion criteria and were coded by year, subject, country, models, and methods. Research activity was modest in 2010–2014, increased from 2015, and peaked in 2021–2022; General Education (n=16) and Program-ming (n=7) dominated subject coverage, while Mathematics, Chemistry, Pedagogy, Eng-lish, and Computer Science appeared infrequently (1–3 occurrences). Publications were concentrated in Asia—India (n=9), China (n=8), Japan (n=3)—with contributions from the United States (n=3) and select European countries. Student Model and Expert Model were the most common abstractions (3 uses each); adaptive learning systems (n=9) and knowledge-graph construction (n=6) were the leading methods. Reported benefits includ-ed individualized pacing and early difficulty detection, whereas recurrent concerns were methodological heterogeneity (12/26) and privacy/ethics (10/26). AI-enhanced education is shifting from heuristic sequencing toward semantically grounded, data-driven orchestra-tion that integrates learner and expert models with adaptive engines and knowledge graphs. To translate technical promise into durable and equitable learning gains, future implementations should broaden subject coverage, standardize reporting to enable me-ta-analysis, rigorously evaluate fairness and privacy, and embed explainable, teach-er-centered workflows across diverse contexts.