Mathematical Explainability Index for Artificial Intelligence Decision Systems in Healthcare and Higher Education
Year: 2026
Type: article
Abstract: Explainable Artificial Intelligence (XAI) is vital for trustworthy decision support in high-impact areas like healthcare and higher education, where transparent AI predictions affect decision quality and stakeholder confidence. Current explainability techniques are often qualitative or model-specific and lack a unified quantitative framework, limiting objective comparison and standardization. This study introduces the Mathematical Explainability Index (MEI), a model-independent quantitative framework that integrates multiple explainability dimensions—transparency, interpretability, fidelity, stability, consistency, complexity, and user trust—into a normalized weighted formula producing a single explainability score. MEI enables direct comparison across machine learning and deep learning models while remaining applicable to diverse application domains. We validate MEI using representative healthcare and higher education datasets, assessing explainability alongside conventional predictive metrics. Comparative experiments reveal MEI’s ability to evaluate models, highlight trade-offs between accuracy and explainability, and guide model selection. Mathematical proofs and sensitivity analyses show the index is consistent, scalable, and robust under varying data and model conditions. Overall, MEI offers a standardized, interpretable quantitative measure to advance trustworthy AI and support transparent deployment of decision support systems in critical sectors.
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
OpenAlex ID: W7225297929