Universities need evidence-based tools to steer digital transformation

Researchers from the Universidad Politécnica Salesiana in Ecuador have developed an interactive analytical framework designed to help universities move digital transformation from broad strategy statements to evidence-based institutional action, using ontology, reliability diagnostics and multiobjective optimization to identify where change is needed and which interventions should be prioritized.

Published in Applied Sciences, the study "Digital Transformation in Higher Education Through Interactive Ontology and Multiobjective Optimization for Evidence-Based Strategic Prioritization" proposes a Python-based Human–Machine Interface that allows higher education institutions to diagnose digital readiness, map the structure of transformation priorities, evaluate measurement consistency and select intervention portfolios under cost, equity, feasibility and readiness constraints.

Universities face pressure to turn digital strategy into measurable action

The authors argue that universities do not become digitally transformed simply by adding platforms, digitizing services or expanding technology infrastructure. Transformation becomes meaningful only when digital capabilities are aligned with governance, quality assurance, academic management, process redesign, institutional intelligence and evidence-based decision-making.

The study particularly focuses on regulated higher education settings, where innovation must coexist with external evaluation, accreditation rules, documentary evidence and internal governance requirements. In such systems, digital transformation has to be traceable, auditable and linked to institutional priorities.

The authors organize the problem around three layers. The first is quality assurance and regulatory alignment, reflecting the need for digital initiatives to support accreditation and external accountability. The second is institutional identity, governance and strategic coherence, recognizing that transformation must fit a university's mission, leadership structures and long-term priorities. The third is process redesign, digital integration, traceability, analytics and operational prioritization, where transformation becomes visible in workflows, information systems and intervention plans.

The study builds on recent research showing that digital transformation depends not only on tools but also on digital skills, faculty capacity, student digital literacy, leadership, organizational support, staff development, data-driven governance, hybrid teaching, curriculum redesign and institutional service improvement. The authors argue that these areas are often studied separately, leaving universities with fragmented evidence and limited support for strategic decision-making.

To address this gap, the proposed framework integrates three analytical layers that are usually treated apart. Ontology-based modeling provides the semantic structure of digital transformation dimensions and their relationships. Statistical reliability analysis evaluates whether grouped indicators behave coherently. Multiobjective optimization helps institutions compare possible intervention portfolios while balancing competing priorities.

The result is not a conventional dashboard that only reports indicators, nor a maturity model that places an institution into a fixed category. The proposed system is built as an interactive decision-support environment where users can filter institutional data, inspect conceptual domains, test reliability, examine readiness patterns and compare strategy options under specific constraints.

Ontology, analytics and optimization form the system's core

The Human–Machine Interface was developed in Python as an analytical environment for institutional diagnosis. It uses a structured local dataset aligned with an open-access digital-readiness reference scenario for higher education institutions, though the authors stress that the data are demonstrative rather than a new survey collected for the study. This gap is critical because the tool is presented as a transferable prototype, not as a fully validated universal model.

The system is organized around seven dimensions of digital transformation readiness: digital infrastructure, digital skills, top management support, readiness for change, e-learning, e-library and e-administration. These dimensions reflect the view that university digital transformation involves infrastructure and services, but also leadership, organizational readiness and academic support systems.

The ontology module turns those dimensions into a structured semantic model. It links the root concept of digital transformation to institutional dimensions and then to item-level indicators. This allows users to see how each observed indicator fits into the wider conceptual framework. The study argues that this semantic structure makes the system more transparent than disconnected indicator dashboards, because each item is anchored to a higher-order institutional construct.

The framework also calculates similarity between dimensions and graph centrality indicators. These outputs help identify which areas occupy a more central role in the transformation model. The study reports that governance-oriented dimensions have strong semantic importance, reinforcing the argument that digital transformation is not mainly a technical infrastructure issue. Leadership, support, strategic alignment and institutional coherence are central to how transformation unfolds.

The statistical module examines measurement consistency. It uses Cronbach's alpha and corrected item-total correlation to test how well indicators contribute to their assigned dimensions. The authors explicitly state that this reliability analysis is a methodological demonstration, not a full empirical validation of the framework. Because the sample was not originally designed to validate their model, the results should be read as internal consistency diagnostics for predefined item groups.

The optimization module is the most directly strategic component. It treats institutional intervention planning as a constrained multiobjective problem. Instead of ranking projects by a single criterion, the system compares intervention portfolios based on cost, readiness gain, equity gain and feasibility. Universities can adjust weights, budget limits, feasibility thresholds and the maximum number of interventions to explore different planning scenarios.

Higher education leaders rarely make digital transformation decisions in ideal conditions. Budget limits, staff capacity, regulatory obligations, equity goals and political feasibility all shape what can realistically be done. The study's optimization model offers a way to identify Pareto-efficient portfolios, meaning sets of interventions where no alternative performs better on one objective without worsening another.

The authors report that no single portfolio dominated all others across readiness, equity, feasibility and composite value. That finding supports their central argument: digital transformation planning should not rely on one-score rankings. It requires trade-off analysis that shows what institutions gain, what they sacrifice and which intervention combinations are efficient under real constraints.

Readiness gaps show need for tailored institutional planning

The results show that digital transformation readiness is uneven across institutional groups and dimensions. Regional profiles were not uniform, and readiness patterns varied across the seven dimensions. Some areas showed closer alignment across groups, while others separated more clearly, especially where managerial support, readiness for change and digital services influenced institutional profiles.

The finding challenges the assumption that a university can be assessed by a single digital maturity score. The authors argue that transformation readiness must be examined through internal distributional patterns, because the same group can appear stronger or weaker depending on whether analysis focuses on lower, median or higher readiness levels. In practical terms, this means institutional leaders need more than headline averages. They need diagnostic tools that show where readiness is uneven and which groups or dimensions need targeted support.

The semantic analysis also shows that digital transformation involves clusters of concepts such as strategy, leadership, support, motivation, interoperability, cybersecurity, connectivity, assessment, artificial intelligence, virtual classes and collaboration. The authors caution that the semantic projection used in the system is exploratory rather than confirmatory, but the output still reinforces a broader point: transformation depends on the interaction of technical, managerial, academic and governance factors.

Investments in platforms, devices or connectivity may fail to produce durable transformation unless they are paired with leadership, training, process redesign, data integration, cybersecurity and quality assurance mechanisms. The framework points toward planning models that connect institutional diagnosis with intervention selection.

The study also highlights equity as a key part of decision-making. Digital transformation in higher education can widen gaps if interventions benefit already prepared groups while neglecting weaker areas. By including equity gain as a criterion in the optimization module, the framework gives universities a way to factor inclusion into strategic planning rather than treating it as a secondary concern.

Feasibility is another major issue identified in the study. Some digital transformation interventions may look attractive in terms of readiness gain but may be difficult to implement because of staff limitations, organizational resistance, costs, governance complexity or regulatory constraints. The model's feasibility threshold is designed to prevent unrealistic options from being included simply because they score well on other objectives.

The authors also draw focus to the framework's limits. The dataset is a structured demonstration sample, not primary institutional data collected for this model. The intervention attributes used in the optimization module are configurable demonstration parameters, not universal empirical coefficients. The Pareto frontiers and portfolio recommendations are therefore illustrative. Real deployment would require recalibration with local cost records, expert panels, institutional performance indicators, budgets and strategic priorities.

The framework is still presented as transferable. The ontology and intervention library can be adapted to different higher education systems. In Latin America, for example, the authors suggest that the ontology could include accreditation, graduate outcomes, research and innovation, social engagement, inclusion, digital governance and evidence traceability. The same architecture could also be adapted to other evidence-heavy domains where organizations need to organize data, interpret complex relationships and prioritize resource-constrained action.

Future work, according to the authors, should validate the framework with primary institutional datasets collected through questionnaires designed specifically for the proposed model. Larger and more diverse samples would allow researchers to test reliability, dimensional validity and cross-institutional stability. The optimization layer also needs expansion to include uncertainty, stricter constraints and scalable solvers for larger intervention libraries.