Work

Decomposing Occupational Strain in Nigerian Universities Using Explainable Machine Learning and Hemodynamic Validation of Workforce Overload and Infrastructural Deficits

Year: 2026
Type: article
Study design: Observational Study
Abstract: Occupational stress in higher education institutions within emerging economies is compounded by acute workforce shortages and chronic infrastructural deficits that resource-constrained academic environments face, but that classical, psychosocial stress models largely overlook. This study isolates and decomposes the two dominant occupational stressors identified in institutional assessment, Facilities and Infrastructure (12.30/20, 61.5 percent saturation) and Workforce and Workload (12.00/20, 60.0 percent), among university staff (N = 32). A multi-stage methodology combines an anonymous, tokenized psychometric protocol with machine learning predictive modelling, explainable artificial intelligence (TreeSHAP and LIME), and clinical cardiovascular cross-validation (n = 5), visualized through six analytical figures spanning dimensional uncertainty, item-level response patterns, feature attribution, clinical correlation, policy comparison, and sensitivity analysis. Clinical triangulation revealed a strong correlation between algorithmic stress scores and Mean Arterial Pressure (r = 0.920, t = 4.07, p = 0.026) and 100 percent diagnostic concordance with independent clinical evaluation. Global SHAP analysis showed that electrical grid collapse and departmental staff shortages jointly explain 79.9 percent of predictive variance, while policy simulation demonstrated that an integrated dual-buffering intervention reduces the institutional stress index by 42.5 percent and burnout prevalence from 56.3 percent to 14.8 percent. The findings are situated against international literature on occupational stress, explainable AI, and algorithmic human resource management, and an evidence-based framework is proposed for sustainable university administration in resource-constrained settings. Given the modest psychometric cohort (N = 32) and especially the small clinical validation subsample (n = 5), these quantitative findings are exploratory and hypothesis-generating, and should be interpreted as preliminary pending larger-scale replication.
Language: English
Field: Psychology
Open Access status: gold
OpenAlex ID: W7225176261
FWCI:
Cites: 44
Cited by:
Related to: 10