Challenges and Opportunities in Higher Education Digital Transformation: A Combined SEM and Machine Learning Perspective

The information age is changing the patterns of higher education institutions across the globe due to the rapid pace of information technology change, emerging governance structures, and rising stakeholder expectations. National policies in Chinese universities have accelerated digitalization, yet the process remains complex as institutional readiness, infrastructure disparities, and incongruent environmental pressures interact in multifaceted ways. To address limitations of earlier models like TAM and UTAUT—which focus mainly on individual user behavior and overlook organizational and environmental forces—this study adopts the Technology-Organization-Environment (TOE) framework combined with Partial Least Squares Structural Equation Modeling (PLS-SEM) and a complementary machine learning workflow to assess the drivers of digital adoption and institutional performance in Chinese HEIs. A cross-sectional, correlational design was employed using data gathered through an online structured questionnaire from 650 respondents, including faculty, students, administrators, and IT personnel from public and private institutions. Measured constructs comprised technological infrastructure, organizational readiness, environmental pressure, digital adoption, and institutional performance. PLS-SEM was used to evaluate measurement and structural models with fit checked using indicators such as composite reliability, AVE, Fornell-Larcker criterion, and SRMR, while machine learning classifiers and clustering techniques were applied to enhance predictive insight and uncover latent adopter profiles. Findings revealed moderate infrastructure and preparedness levels (Internet connectivity M = 3.08, staff training M = 2.89). Hypothesis testing identified statistically insignificant relationships along the main paths: Technological Infrastructure → Digital Adoption (β = − 0.064, p = 0.104), Organizational Readiness → Digital Adoption (β = − 0.006, p = 0.879), Environmental Pressure → Digital Adoption (β = 0.065, p = 0.097), and Digital Adoption → Institutional Performance (β = 0.000, p = 0.993), with poor model fit suggesting low predictive power. Machine learning analyses provided additional perspectives, highlighting nuanced behavioral segments and classification patterns beyond those revealed by SEM. While the TOE framework remains conceptually robust, these findings emphasize the value of integrating advanced analytics and suggest that unmeasured factors—such as leadership culture, stakeholder engagement, or digital strategy alignment—may play critical moderating roles. An integrated strategy combining digital leadership, capacity development, inter-institutional collaboration, and data-driven predictive modeling is essential to unlock the transformative potential of digital technologies in Chinese HEIs.