Work

A hybrid AHP–PID framework balancing structural stability and behavioral adaptivity for 5G-enabled innovation and entrepreneurship education

Year: 2026

Type: article

Abstract: Ultra-low latency and high-frequency interaction in 5G-enabled learning environments pose engineering challenges to quality evaluation, as existing static and dynamic models cannot effectively balance structural stability and real-time responsiveness. This challenge is particularly pronounced in innovation and entrepreneurship education (IEE), where intensive interaction, rapid feedback, and heterogeneous learning behaviors are integral to instructional effectiveness. This study develops and validates a hybrid AHP–PID dynamic evaluation framework grounded in an integrated Structure–Behavior–Regulation perspective that synthesizes Complex Systems Theory (CST), the Theory of Planned Behavior (TPB), and Digital Control Theory (DCT) to reconcile structural stability and behavioral adaptivity. Based on a rigorously constructed 4–12–36 hierarchical indicator system incorporating pedagogical, behavioral, and 5G-specific metrics (e.g., 5G-supported AR/VR immersion and network latency), the framework uses high-frequency behavioral observations collected in 5G-enabled environments to support periodic re-weighting of the evaluation indicator set in an open-loop scoring design that does not actuate pedagogical intervention, while preserving structural consistency through expert-derived static weights. Using data from 200 participants (100 teachers and 100 students) across comprehensive universities, applied undergraduate institutions, and vocational colleges, the hybrid model achieves higher predictive accuracy than the traditional static Analytic Hierarchy Process (AHP), standalone dynamic models, and machine-learning benchmarks on the present cohort. It achieves an AUC of 0.94 and a Pearson r of 0.76 against an expert-specified multi-source gold standard. The results further indicate that an empirically optimized fusion coefficient β of 0.65 underpins these performance gains by effectively balancing institutional comparability and behavioral adaptivity, while exploration subgroup analysis across institutional types and course categories revealed no significant performance degradation ( p = 0.18). The comprehensive‑university subsample ( n = 34, 17%) is underpowered, so these subgroup generalizability results should be regarded as preliminary. Theoretically, this study advances a coherent Structure–Behavior–Regulation logic that clarifies how systemic structure, learner behavior, and adaptive regulation interact in digital educational evaluation. Practically, the framework supports a shift from post hoc assessment toward timely quality governance and offers a replicable and adaptive evaluation template for 5G-enabled IEE and similar technology-intensive learning environments.

Source: Scientific Reports

DOI: 10.1038/s41598-026-74665-3

Authors: Chuanqi Cheng, Mazita Binti Mokhtar, Zarith Sufia Binti Azlan, Jinxia Huang

Institutions: Universiti Malaysia Pahang Al-Sultan Abdullah, Guangxi Science & Technology Normal University

Language: English

Topic: Educational Technology and Pedagogy

Subfield: Education

Field: Social Sciences

Domain: Social Sciences

SDGs: Quality education, Industry, innovation and infrastructure

Open Access status: gold

Estimated APC paid: $2,690