Abstract
{ "background": "Public health surveillance systems are critical for disease control, yet their reliability and diagnostic performance in low-resource settings are often inadequately quantified. Existing evaluations are frequently cross-sectional and lack robust methods to account for hierarchical data structures and temporal variability.", "purpose and objectives": "This study aims to develop and apply a novel longitudinal Bayesian hierarchical model to evaluate the reliability and diagnostic accuracy of integrated disease surveillance and response (IDSR) systems in a sub-Saharan African context over an extended period.", "methodology": "We conducted a longitudinal analysis of national surveillance data. The core statistical model is a Bayesian hierarchical latent class model: $y{itk} \\sim \\text{Bernoulli}(\\pi{itk})$, $\\text{logit}(\\pi{itk}) = \\alphak + \\beta{tk} + u{ik} + \\epsilon{itk}$, where $y{itk}$ is the reported incidence for disease $k$ in region $i$ at time $t$. Parameters were estimated using Hamiltonian Monte Carlo, with posterior credible intervals used for inference.", "findings": "The analysis indicates substantial temporal heterogeneity in system sensitivity, with a posterior probability of 0.92 that sensitivity for priority notifiable diseases declined in the latter half of the study period. Model estimates show the median system sensitivity across all diseases was 0.65 (95% CrI: 0.58, 0.71), highlighting significant under-reporting.", "conclusion": "The proposed model provides a robust framework for longitudinal surveillance evaluation, revealing critical, time-varying weaknesses in system performance that cross-sectional analyses fail to detect.", "recommendations": "Implement routine longitudinal reliability assessments using hierarchical modelling. Allocate resources to strengthen data quality in consistently low-performing geographical and disease-specific subsystems.", "key words": "Bayesian hierarchical model, disease surveillance, diagnostic accuracy, longitudinal analysis, public health informatics, system reliability", "contribution statement": "This paper introduces a novel longitudinal