AI Boosts Teaching Quality in Yemen’s Health Education Institutions, Study Finds

In one of the first empirical studies of artificial intelligence adoption in a crisis-affected higher education system, researchers have found that AI technologies are significantly associated with better teaching quality in Yemen’s university-level health education institutions. The study, published in Discover Education, surveyed 280 faculty members and academic leaders across five universities in Sana city and found a strong, statistically significant positive relationship between the use of AI tools and the perceived quality of the educational process. The finding matters far beyond Yemen: it offers rare, hard data on how digital transformation can function in settings where electricity is unreliable, bandwidth is scarce, and budgets are stretched to breaking.

The research, conducted by Ammar Ali Al-Raimi of Al-Razi University in Sana’a, adopted a descriptive analytical design grounded in established technology adoption theory. The study drew on the DeLone and McLean Information Systems Success Model and the Technology Acceptance Model to frame its hypothesis that AI functions as enabling infrastructure for pedagogical quality. Data were collected between January and June 2025 using a structured questionnaire administered both electronically and on paper, with a response rate of 87.5 percent. The sample was drawn from a population of roughly 1,001 academic staff across colleges of medicine, pharmacy, nursing, and allied health sciences at both public and private universities, using stratified random sampling to ensure proportional representation.

The analytical approach was rigorous. The researcher employed partial least squares structural equation modeling, or PLS-SEM, a technique well suited to exploratory, prediction-oriented research with complex multidimensional constructs and moderate sample sizes. Bootstrapping with 5,000 resamples generated path coefficients, t values, and p values, while a battery of validity checks — indicator loadings, Cronbach’s alpha, composite reliability, average variance extracted, the Fornell-Larcker criterion, and heterotrait-monotrait ratios — confirmed that the measurement model was sound. Common method bias, a persistent risk in single-survey studies, was assessed twice: Harman’s single-factor test extracted only 35 percent of variance, below the 40 percent threshold, and Kock’s full collinearity assessment returned variance inflation factors well under the conservative limit of 3.3.

The headline result was striking. The structural model showed a path coefficient of 0.79 between AI technology use and educational process quality, with a t value of 24.10 and a p value below 0.001 — an exceptionally robust association. The model explained 62 percent of the variance in perceived educational quality, with a Q² of 0.41 confirming strong predictive relevance and fit indices, including an SRMR of 0.067 and an NFI of 0.912, indicating satisfactory model fit. In plain terms: where faculty reported heavier use of AI tools, they also reported markedly better teaching, assessment, and learning outcomes.

The study broke AI adoption into four dimensions, and the differences between them proved revealing. Profiling and prediction — the use of learning analytics to identify at-risk students and support academic decision-making — emerged as the strongest predictor, with a path coefficient of 0.34. Personalized learning followed at 0.29, and intelligent assessment at 0.26. Intelligent tutoring and support systems, such as virtual conversational assistants and AI-driven academic advising, showed the weakest but still significant effect at 0.18. The reason for that gap is deeply contextual: interactive tutoring systems demand continuous real-time connectivity, high server capacity, and sophisticated instructional design, all of which are scarce in an environment marked by recurrent power outages and constrained internet bandwidth, and by the absence of Arabic-language, medically tailored conversational platforms.

Baseline levels told their own story. Overall AI utilization registered a mean score of 3.12 on a five-point scale — moderate, at 62.4 percent — indicating that adoption has begun but remains far from fully integrated. Educational process quality scored similarly at 3.25, or 65 percent, suggesting institutions perceived as acceptable but with substantial room for improvement. Within the AI dimensions, profiling and prediction led at a mean of 3.28, while intelligent tutoring trailed at 2.94. Among quality dimensions, good teaching scored highest at 3.34, followed by learning outcomes at 3.27. The pattern suggests a system leaning on low-cost, asynchronous analytics rather than bandwidth-hungry interactive tools — a rational adaptation to infrastructure constraints.

The barriers data may be the study’s most consequential contribution. Weak technological infrastructure topped the list at a high level, with a mean of 3.78, followed closely by the shortage of specialized AI training programs at 3.65. Thematic categorization of the barriers placed infrastructure limitations at 75.6 percent intensity and faculty training deficits at 73.0 percent, with governance, financial, and ethical concerns representing moderate institutional challenges. The absence of clear institutional policies on AI use compounded these problems. In other words, the constraint is not skepticism or lack of interest among educators — it is plumbing, training, and policy.

Equally notable was what did and did not predict positive perceptions. Faculty members’ digital experience significantly shaped their views of AI’s value, with a one-way ANOVA returning F = 4.39 and p = 0.013, and participation in AI-related training programs produced a significant difference as well, with t = 2.32 and p = 0.021. Academic qualifications, by contrast, made no significant difference, with F = 1.26 and p = 0.284. Practical exposure and targeted professional development, not formal degrees, determine who embraces educational AI — a finding with direct implications for how ministries and universities should spend scarce capacity-building funds.

From these findings, the study proposes a phased, pragmatic implementation strategy tailored to resource-constrained environments. Phase one prioritizes low-bandwidth, open-access solutions: automated grading assistance, offline learning management systems, and open-source analytics for performance tracking. Phase two establishes modular faculty training workshops covering AI literacy, prompt engineering, and ethical usage — the intervention the data show moves perceptions most. Phase three calls for policy and infrastructure investment, including transparent guidelines on student data privacy and academic integrity, and secure digital architectures such as cloud-hosted platforms. The study also outlines a five-component strategic framework spanning technological infrastructure, human resource development, pedagogical AI integration, institutional governance, and continuous quality evaluation, positioned as an operational blueprint for academic leaders rather than a statistically tested model.

The study’s limitations are candidly acknowledged: the data are self-reported, the sample is confined to faculty and leaders in Sana city, and the cross-sectional design cannot establish causation over time. The author calls for longitudinal and mixed-method research across diverse regions, incorporating student perspectives and classroom observation. Still, the contribution is substantial. By quantifying the relationship between AI adoption and educational quality in one of the world’s most challenging educational environments, the study demonstrates that the AI-in-education revolution is not a luxury of wealthy institutions. Even under severe constraint, the evidence suggests, the right low-cost tools — analytics first, tutoring later — can measurably strengthen the training of the healthcare professionals a health system most depends on.

Subject of Research: The role of artificial intelligence technologies in enhancing educational process quality in university health education institutions in Yemen

Article Title: Employing artificial intelligence technologies to enhance educational process quality in health education institutions in Yemen

Article References: Al-Raimi, A. A. (2026). Employing artificial intelligence technologies to enhance educational process quality in health education institutions in Yemen. Discover Education, 5(1), Article 861. https://doi.org/10.1007/s44217-026-02133-1

Image Credits: AI Generated

DOI: 10.1007/s44217-026-02133-1

Keywords: artificial intelligence, higher education, health education, Yemen, educational quality, PLS-SEM, learning analytics, personalized learning, intelligent tutoring, faculty training, developing countries, educational technology