Longitudinal Evaluation of Public Health Surveillance Methodologies in Nigeria: A Time-Series Forecasting Model for Risk Reduction, 2000–2026

{ "background": "Public health surveillance systems in Nigeria have historically relied on lagged, aggregate data, limiting proactive risk management. A critical methodological gap exists in evaluating these systems' predictive performance for forecasting disease burdens and measuring the efficacy of interventions over extended periods. ", "purpose and objectives": "This study aimed to develop and validate a time-series forecasting model to evaluate the methodological performance of national surveillance systems in measuring longitudinal risk reduction for key communicable diseases. ", "methodology": "A longitudinal study design was employed, analysing nationally reported surveillance data. We developed a Seasonal Autoregressive Integrated Moving Average with exogenous variables (SARIMAX) model, specified as \ (B) \ (Bˢ) \ᵈ\D yt = \ (B) \ (Bˢ) \ + \ Xt, to generate forecasts. Model performance was assessed using rolling-origin evaluation and compared against observed data to quantify predictive accuracy and bias. ", "findings": "The forecasting model demonstrated a mean absolute percentage error (MAPE) of 12. 3% (95% CI: 10. 8, 13. 9) for out-of-sample predictions across three major disease groups. A key finding was a systematic overestimation of reported malaria incidence by the model in the latter half of the study period, suggesting a potential surveillance artefact or genuine reduction in transmission not fully captured by baseline covariates. ", "conclusion": "The SARIMAX model provides a robust methodological framework for the longitudinal evaluation of surveillance data, revealing systematic forecasting errors that indicate possible improvements in disease control or limitations in current reporting mechanisms. ", "recommendations": "We recommend the integration of this forecasting methodology into routine surveillance system evaluations to identify anomalies and improve data quality. Future models should incorporate higher-resolution spatial and behavioural data. ", "key words": "surveillance evaluation, forecasting, time-series analysis, risk assessment, public health, Nigeria", "contribution statement": "This paper provides a novel application of the SARIMAX model for