Abstract
<jats:p>Artificial intelligence has substantially improved infectious disease surveillance, yet most explainable AI (XAI) techniques remain limited to statistical feature attribution, providing little epidemiological context to support public-health decision-making. This study proposes a semantic-based explainable artificial intelligence framework that bridges the semantic gap between machine-learning predictions and biologically grounded interpretation through ontology-guided reasoning. The framework integrates stochastic prediction models (XGBoost, Gaussian Process Regression, and Elastic Net), SHAP-based feature interrogation, semantic concept mapping, OWL/SWRL reasoning, and natural-language explanation synthesis to transform statistical associations into epidemiologically plausible reasoning traces. The framework was evaluated using a longitudinal malaria surveillance dataset comprising 6,300 spatio-temporal observations collected from 30 health centres across six states in North-Eastern Nigeria over seven months. Model performance was assessed using a harmonized chronological holdout and Leave-One-District-Out Spatial Cross-Validation protocol, while semantic validity was evaluated through ontology alignment, expert assessment, and statistical validation. Elastic Net achieved the strongest predictive performance R2=0.901, whereas XGBoost produced the most stable explanations (SHAP rank correlation = 0.87). The semantic layer achieved 92.3% ontology coverage, 94.6% mapping accuracy, and a semantic fidelity score of 0.91. Independent evaluation by seven infectious disease specialists demonstrated significant improvements in semantic causability, clinical actionability, and forensic trust over conventional XAI approaches (Wilcoxon, (p &amp;lt; 0.01); effect size (r = 0.71)). These findings demonstrate that integrating semantic reasoning with machine learning produces transparent, context-aware, and epidemiologically coherent explanations, providing a practical neuro-symbolic framework for trustworthy AI-enabled infectious disease surveillance.</jats:p>